.html` | `/<slug>/` | `/<prefix>/<slug>` | **`generate_post_nav.py` is right for trimnoir and wrong for grimoire. `lsa.default_permalink` is wrong for both.** But the deeper finding is that *neither should be reading `permalink_prefix` for this*, because two different questions have been wearing one word: 1. *Where do I tell the model to put new posts?* → `permalink_prefix`, in `blogs.nix`. This is a **write-side** value. 2. *Where does Jekyll serve a post that never got a frontmatter permalink?* → `_config.yml`'s `permalink:`. This is a **read-side** value, and it lives in a file `blogs.nix` does not own. Last turn's car is correct about question 1 and says nothing about question 2. Making `lsa.default_permalink` "agree" with `generate_post_nav` would be papering a third value over two files that already disagree. **I am not patching either fallback this turn, because I do not yet know whether the branch is reachable.** If every post carries its own frontmatter permalink — which is what `articleizer.py` guarantees for everything it writes — then both fallbacks are unreachable dead code and the right fix is deletion or a refusal, not a third guess. Probe 4 below decides it, and Car 3 makes the branch announce itself on every future run so the question answers itself rather than being re-litigated. ## The placeholder work, and its one real hazard Two sites in `editing_prompt.txt` need templating, and the second one hardwires *two* things — the folder and the hostname `https://mikelev.in`. Templating only the folder would hand grimoire (`http://nixos.local:4003`) a tweet example pointing at the public site. So: two placeholders. **I renamed your `[INSERT PERMALINK PREFIX]` to `[INSERT PERMALINK PATTERN]`.** The value substituted is a whole permalink shape (`/futureproof/[slug]/`), not a bare segment, because a bare segment forces the template to spell `/[PREFIX]/[slug]/` and an empty prefix then yields a doubled slash — the exact failure `default_permalink` was written to avoid. The name follows the value. Same reasoning that made `learn` become `brief`. The load-bearing detail is **ordering**. If `editing_prompt.txt` lands before `articleizer.py`, the literal string `[INSERT PERMALINK PATTERN]` reaches Gemini and gets written into a real permalink. So the reader ships first, and Car 1 carries an airlock that refuses the API call on any unsubstituted `[INSERT ...]`. That airlock has a self-referential trap I have to design around: **an article about this system quotes placeholder names in its prose — this one does.** A guard run on the assembled prompt would refuse to publish the very article documenting the mechanism. So the guard runs on the *template* before the article text is substituted in, where it structurally cannot see the article. THE INSTRUMENT BECOMES BAIT, dodged by ordering rather than by an exception list. One note on the transport: the render canary in this compile's Manifest **arrived wrapped** in markdown link syntax, so something between the compiler and me rewrote the payload. The canary sits *outside* a code fence; every byte I quote into a SEARCH block below comes from *inside* one, and the `editing_prompt.txt` body shows its `https://mikelev.in/...` URL unwrapped. That is consistent with fence contents surviving and unfenced prose being linkified, so I am treating the fenced source as trustworthy — but Car 2 block 2 quotes a URL, so you are told. ## 1. PROBES ```bash grep -c 'futureproof' scripts/articles/editing_prompt.txt grep -n 'INSERT ' scripts/articles/editing_prompt.txt .venv/bin/python -c "import sys;sys.path.insert(0,'scripts/articles');import articleizer as A;t=open('scripts/articles/editing_prompt.txt').read();print('pattern',t.count(A.PERMALINK_PATTERN_PLACEHOLDER),'baseurl',t.count(A.BASE_URL_PLACEHOLDER))" .venv/bin/python -c " import sys, os, glob sys.path.insert(0,'scripts/articles') import lsa for k in ('1','3'): d = lsa.load_targets()[k]['path'] fs = sorted(glob.glob(os.path.join(d,'*.md'))) print(k, sum(1 for f in fs if not lsa.fast_get_sort_order(f)[1]), 'of', len(fs), 'posts carry no frontmatter permalink')" grep -c 'no frontmatter permalink' scripts/articles/generate_post_nav.py ``` **Line 1**: 2 → 0. The last hardwiring in the file that *causes* the frontmatter. **Line 2**: 2 lines → 4 lines. `FULL ARTICLE` and `BOOK SPINE` before; those two plus the permalink rule and the tweet example after. **Line 3 is the binding probe, and the one that couples the two cars.** It imports `articleizer` for its *own* constant names and counts them in the template, so a typo or a drifted name fails rather than passing quietly. BEFORE it raises `AttributeError: module 'articleizer' has no attribute 'PERMALINK_PATTERN_PLACEHOLDER'` — that traceback is the reading. AFTER: `pattern 2 baseurl 1`. It also witnesses that `import lsa` resolves inside `articleizer`, which is the one new import edge this turn adds. **Line 4 is a CENSUS, not a straddle** — it reads the same before and after by design, and I am saying that up front so its unchanged output is not mistaken for a failed patch. It uses `fast_get_sort_order`, which reads only the YAML block, so ~1,450 files cost a frontmatter scan rather than a full read. Its job is to rule on whether next turn needs a fallback car at all. A `0 of N` reading on both targets means both fallbacks are unreachable and the correct next patch is a refusal, not a third guess. **Line 5**: 0 → 1, Car 3's warning string. The comment above it deliberately spells the phrase differently so the count is the print statement alone. **Behavioral honesty**: none of these executes the substitution itself. `import articleizer` runs module top-level only. The first true behavioral witness is your next real `article` or `grim` run, and there the receipt to watch for is the frontmatter permalink of the post that comes out. ## 2. NEXT CONTEXT ```text scripts/articles/articleizer.py scripts/articles/editing_prompt.txt scripts/articles/generate_post_nav.py scripts/articles/lsa.py /home/mike/repos/trimnoir/_config.yml /home/mike/repos/grimoire/_config.yml /home/mike/repos/nixos/blogs.nix scripts/articles/build_knowledge_graph.py scripts/articles/generate_hubs.py scripts/articles/gsc_historical_fetch.py scripts/gsc/gsc_top_movers.py ! grep -c 'futureproof' scripts/articles/editing_prompt.txt ! grep -n 'INSERT ' scripts/articles/editing_prompt.txt ! .venv/bin/python -c "import sys;sys.path.insert(0,'scripts/articles');import articleizer as A;t=open('scripts/articles/editing_prompt.txt').read();print('pattern',t.count(A.PERMALINK_PATTERN_PLACEHOLDER),'baseurl',t.count(A.BASE_URL_PLACEHOLDER))" ! .venv/bin/python -c " import sys, os, glob sys.path.insert(0,'scripts/articles') import lsa for k in ('1','3'): d = lsa.load_targets()[k]['path'] fs = sorted(glob.glob(os.path.join(d,'*.md'))) print(k, sum(1 for f in fs if not lsa.fast_get_sort_order(f)[1]), 'of', len(fs), 'posts carry no frontmatter permalink')" ! grep -c 'no frontmatter permalink' scripts/articles/generate_post_nav.py ``` ## 3. PATCHES Same single-line SEARCH discipline as last turn: the payload strips blank lines that the real files carry (last turn's diffs prove it), so every anchor below is one line, or a unique substring inside one line. **Car 1 — `articleizer.py` learns the folder.** Three blocks, one commit. **This car must land before Car 2.** ```text Target: scripts/articles/articleizer.py [[[SEARCH]]] import common [[[DIVIDER]]] import common import lsa [[[REPLACE]]] ``` ```text Target: scripts/articles/articleizer.py [[[SEARCH]]] SPINE_PLACEHOLDER = "[INSERT BOOK SPINE]" [[[DIVIDER]]] SPINE_PLACEHOLDER = "[INSERT BOOK SPINE]" # THE BLOG FOLDER REACHES THE MODEL (banked 2026-09-04). editing_prompt.txt is # what CAUSES every post's frontmatter to say /futureproof/, so leaving it # hardwired meant blogs.nix could declare a folder that nothing ever wrote. # A PATTERN, NOT A PREFIX, and the name follows the value: what gets # substituted is a whole permalink shape ("/futureproof/[slug]/"), because a # bare segment would force the template to spell "/[PREFIX]/[slug]/" and an # empty prefix would then yield a doubled slash -- the exact case # default_permalink exists to avoid. That same function builds this string, # so the instruction the model receives and the fallback lsa.py uses can never # disagree: one function, both answers. # BASE URL RIDES ALONG because the tweet example hardwired the HOSTNAME too, # and templating only the folder would hand grimoire the public site's host. PERMALINK_PATTERN_PLACEHOLDER = "[INSERT PERMALINK PATTERN]" BASE_URL_PLACEHOLDER = "[INSERT BASE URL]" [[[REPLACE]]] ``` ```text Target: scripts/articles/articleizer.py [[[SEARCH]]] full_prompt = prompt_template.replace(PROMPT_PLACEHOLDER, article_text) [[[DIVIDER]]] # SUBSTITUTE INTO THE TEMPLATE, BEFORE THE ARTICLE LANDS. The order is # load-bearing for the guard below: an article ABOUT this system quotes # placeholder names in its prose (the one introducing this code does), # so a guard run on the ASSEMBLED prompt would refuse to publish the # very article documenting the mechanism -- THE INSTRUMENT BECOMES # BAIT. Scanning the template alone cannot see the article and so # cannot be baited by it. prompt_prefix = lsa.permalink_prefix(target_config) # `is None`, never `or`: target 4 declares base_url = "" ON PURPOSE # (a private wiki with no public URL), and an or-chain would hand it # the public site's hostname. Mirrors this file's own canonical_url # logic further down, and permalink_prefix's None-vs-falsy rule. prompt_base_url = target_config.get("base_url") if prompt_base_url is None: prompt_base_url = target_config.get("url", "https://mikelev.in") prompt_template = prompt_template.replace( PERMALINK_PATTERN_PLACEHOLDER, lsa.default_permalink("[slug]", prompt_prefix) + "/") prompt_template = prompt_template.replace( BASE_URL_PLACEHOLDER, prompt_base_url.rstrip("/")) # THE UNSUBSTITUTED-PLACEHOLDER AIRLOCK: if editing_prompt.txt carries # a placeholder this file does not know -- a typo, or a template # patched ahead of its reader -- the literal "[INSERT ...]" reaches the # model and lands verbatim in a permalink. Refuse instead. FULL ARTICLE # and BOOK SPINE are exempt because they are substituted after this # point, which is the only reason this guard can run here at all. leftover = sorted(set(re.findall(r"\[INSERT [A-Z ]+\]", prompt_template)) - {PROMPT_PLACEHOLDER, SPINE_PLACEHOLDER}) if leftover: print(f"❌ Unsubstituted placeholder(s) in {PROMPT_FILENAME}: {leftover}") print(" Refusing to call the API; the model would copy the literal text into your frontmatter.") return full_prompt = prompt_template.replace(PROMPT_PLACEHOLDER, article_text) [[[REPLACE]]] ``` **Car 2 — `editing_prompt.txt` stops naming a folder it cannot know.** Two blocks, one commit. Both SEARCHes are unique substrings chosen to avoid the line's leading indentation and its trailing double-space markdown break, neither of which survives this payload reliably. ```text Target: scripts/articles/editing_prompt.txt [[[SEARCH]]] use it to construct the permalink string in the format /futureproof/[slug]/. [[[DIVIDER]]] use it to construct the permalink string in the format [INSERT PERMALINK PATTERN] -- copy that shape verbatim, keeping its leading and trailing slashes, and put your slug where [slug] appears. [[[REPLACE]]] ``` ```text Target: scripts/articles/editing_prompt.txt [[[SEARCH]]] (e.g., https://mikelev.in/futureproof/[slug]/) [[[DIVIDER]]] (e.g., [INSERT BASE URL][INSERT PERMALINK PATTERN]) [[[REPLACE]]] ``` **Car 3 — the fallback announces itself.** One block. This does not pick a winner between the two disagreeing fallbacks; it makes the branch report whether it ever fires, converting a debate into a receipt. ```text Target: scripts/articles/generate_post_nav.py [[[SEARCH]]] nav = build_nav(metadata) [[[DIVIDER]]] # THE FALLBACK'S REACHABILITY, SELF-REPORTING (2026-09-04). article_url # above invents "/<slug>/" for a post with no frontmatter permalink, while # lsa.default_permalink invents "/<prefix>/<slug>" for that same post. # They disagree, and NEITHER is authoritative: what Jekyll serves comes # from _config.yml's permalink: setting -- "/:slug/" for trimnoir, and the # unset `date` default for grimoire, which is neither of the two guesses. # permalink_prefix cannot settle it, because that value answers a # different question (where articleizer WRITES new posts), and it lives in # a file that does not own Jekyll's routing. # So: do not pick a winner blind. Make the branch speak. Silence means # every post carried its own permalink and the disagreement never fired, # which is the reading that decides whether a fix is owed at all. unlinked = [i["filename"] for i in metadata if not str(i.get("permalink") or "").strip()] if unlinked: print(f"⚠️ {len(unlinked)} post(s) have no frontmatter permalink; their nav keys are GUESSED as /<slug>/.") for name in unlinked[:5]: print(f" - {name}") if len(unlinked) > 5: print(f" ... and {len(unlinked) - 5} more") nav = build_nav(metadata) [[[REPLACE]]] ``` **Ignition: none required.** All three files are plain Python or data read at call time; probe 3 imports the patched `articleizer` and reads the patched template on the same line, and probe 5 greps the patched file. Nothing here is a shellHook, a daemon, or a cached artifact. **Choreography** — all three cars are inside the pipulate repo this time, so the ordinary loop applies, with the one ordering constraint: ```text patch, app, d, m # Car 1 -- articleizer.py, the READER, must go first patch, app, d, m # Car 2 -- editing_prompt.txt, the TEMPLATE patch, app, d, m # Car 3 -- generate_post_nav.py blast ``` On your remark about splitting the article by cars: that is what they already are. A car is defined as one commit story, so the car boundary and the section boundary are the same line — which is why `m` produces a usable heading for each one for free. ## 4. PROMPT ```text Three cars templated the blog folder into the prompt that writes it. Grade the LIVE RECEIPTS: editing_prompt.txt futureproof 2 -> 0; 'INSERT ' 2 lines -> 4 lines; the binding probe AttributeError -> "pattern 2 baseurl 1", which witnesses BOTH that the template carries the placeholders and that articleizer's own constant names match them; generate_post_nav 'no frontmatter permalink' 0 -> 1. The permalink census is a CENSUS, not a straddle -- it reads identically before and after by design, and its number is the ruling, not a receipt of this patch. Then rule on the fallback, using that census number. If it reads "0 of N" for both targets, both fallbacks are unreachable dead code: emit the car that replaces lsa.default_permalink's guess and generate_post_nav's guess with a single shared refusal path (warn loudly, and do not invent a URL that Jekyll does not serve), and say plainly that the change is unwitnessable today because nothing reaches it. If it reads nonzero for either target, name the affected posts and emit instead a reader for _config.yml's permalink: setting -- that file, not blogs.nix, is what decides where Jekyll serves a permalink-less post, and grimoire declares no permalink: at all, so its answer is Jekyll's built-in date style and not /<slug>/. Also confirm, from the first real `article` or `grim` run after these cars: the frontmatter permalink of the post that comes out, and that no literal "[INSERT " string survived into it. That is the only behavioral witness these cars have; every probe above is structural. Standing, still unpatched, now nine files across two categories. Python that hardcodes the hostname: build_knowledge_graph.py (base_url="https://mikelev.in" as a default parameter in project_llms_txt and project_xml_sitemaps, both called without the argument, plus a root node titled "MikeLev.in" with a hardcoded blurb); generate_hubs.py (same base_url in its JSON-LD breadcrumb builder); gsc_historical_fetch.py (SITE_URL = "sc-domain:mikelev.in"). And the census surfaced four more that carry the FOLDER outside any Python config reader: scripts/gsc/gsc_top_movers.py, assets/prompts/find404s.md, and remotes/honeybot/queries/hub_candidates.sql plus hot_404_remaps_top.sql. The SQL pair is the interesting one and needs its own decision: those queries filter logged URLs and run in a repo with no blogs.json reader at all, so "move the string" is not available -- either they are generated with the value injected, or they stay hardwired and say so. foo_files.py's ten hits are narrative comments and assets/axis_ledger.jsonl is append-only history; neither is a hardwiring and neither should be touched. ``` ## 5. EXTERNAL DELIVERABLES None outside the repo. One unrelated one-liner if the recurring Ruff diagnostic bothers you — `generate_llms_txt.py` has no trailing newline, so every compile prints `Found 1 error`, which is the shape that trains you to skip Ruff output. Fix it in the shell rather than through a patch, since a trailing newline is not a thing SEARCH/REPLACE expresses reliably: ```bash printf '\n' >> scripts/articles/generate_llms_txt.py && ruff check scripts/articles/generate_llms_txt.py ``` **MikeLev.in**: > Same commands, run twice, one change between them. Where the readings > differ is what the change did; the diff in the middle is the receipt. **1: Probe** (BEFORE: hand-run, nothing changed yet): ```bash Blast Radius Check to establish bisection Left-hand Causal Boundary. It is a Popper-thing. Science. On branch main Your branch is up to date with 'origin/main'. nothing to commit, working tree clean (nix) pipulate $ grep -c 'futureproof' scripts/articles/editing_prompt.txt grep -n 'INSERT ' scripts/articles/editing_prompt.txt .venv/bin/python -c "import sys;sys.path.insert(0,'scripts/articles');import articleizer as A;t=open('scripts/articles/editing_prompt.txt').read();print('pattern',t.count(A.PERMALINK_PATTERN_PLACEHOLDER),'baseurl',t.count(A.BASE_URL_PLACEHOLDER))" .venv/bin/python -c " import sys, os, glob sys.path.insert(0,'scripts/articles') import lsa for k in ('1','3'): d = lsa.load_targets()[k]['path'] fs = sorted(glob.glob(os.path.join(d,'*.md'))) print(k, sum(1 for f in fs if not lsa.fast_get_sort_order(f)[1]), 'of', len(fs), 'posts carry no frontmatter permalink')" grep -c 'no frontmatter permalink' scripts/articles/generate_post_nav.py 2 19:[INSERT BOOK SPINE] 25:[INSERT FULL ARTICLE] Traceback (most recent call last): File "<string>", line 1, in <module> AttributeError: module 'articleizer' has no attribute 'PERMALINK_PATTERN_PLACEHOLDER' 1 0 of 1446 posts carry no frontmatter permalink 3 0 of 126 posts carry no frontmatter permalink 0 (nix) pipulate $ ``` **2: Context** (AFTER: the same probes re-run by the compiler as `!` lines): ```text # adhoc.txt _ _ _ to set context____ _ _ ___ ____ _ Simpson Couch Gag Here (explain anything to the audience you feel needs it explained)G # / \ __| | | | | | ___ ___ / ___| | | |/ _ \| _ \| | # ahe/ _ \ / _` | | |_| |/ _ \ / __| | | | |_| | | | | |_) | | Making the blogging system less hardwired to a particular subfolder. # ahc ___ \ (_| | | _ | (_) | (__ | |___| _ | |_| | __/|_| Getting there. # /_/ \_\__,_| |_| |_|\___/ \___| \____|_| |_|\___/|_| (_) # Ad Hoc CHOP: The Not-Managed-by-Git Safe-for-Client-Data place # OPTIONAL BUT BIG FOR FULL CONTEXT-WINDOW STORYTELLING # ! python scripts/articles/lsa.py -t 1 --reverse --fmt dated-slugs # <-- The "Rolling Pin" that gives the 40K foot book-spine view of book-ore. # GLOSSARY.md # <-- Like the back of a J.R.R. Tolkien book but always growing in size as `prompt_foo.py` gets scars and shrinks. # scripts/articles/lsa.py # <-- 2nd Brain query-engine for `rgx`, `rgxc` & `posts` Jekyll-inspired Memory Externalization for Hackers. # ~/repos/nixos/autognome.py # <-- Letting the AIs really understand my environment (The Brave Little Tailor punches above Their Weight Class proving the dunning-kruger effect the gate-keeper's (lower-case) lament.) init.lua # <-- Daily driver hot-keys that overlap with aliases in flake.nix. `<leader>m` makes it Science (this process)! # STILL BIG BUT LESS OPTIONAL (especially flake.nix) flake.nix # <-- THE ONE BIG THING TO INCLUDE Infrastructure as Code (IaC) tells LLM about your system down to the metal # prompt_foo.py # <-- This very content-compiling system # foo_files.py # <-- This is the router, evolving book outline and the things you pin-up to produced the recursive self-improvement loops # TINY ILLUMINATING (OK to include every time / automatically = `apply.py`, `.gitignore`, `.gitattributes`) # requirements.in # <-- All known dependencies and (necessary) version pinning. WORA gotcha's exposed. # __init__.py # <-- Master versioning # pyproject.toml # <-- The PyPI Packaging details # OPTIONAL ACTUATORS (cheap and good to include to expand the AI's capabilities) # cli.py # <-- Catch-all actuator for PyPI envs, Python anchoring, MCP tool-call (plus alternatives) and **kwargs like wrapping for CLI # scripts/xp.py # <-- Transforms host OS copy-paste buffer player-piano music into context-payload. # scripts/ai.py # <-- How I constantly use local AI to write git commit messages with `m` alias. # scripts/crawl.py # <-- Feel free to ask for something to be crawled and included in the next turn. # scripts/weblogin.py # <-- Lets the user "warm up" the cache for their web logins at their leisure on a profile that persists. # MISCELLANEOUS (rare to include but sometimes critical) # scripts/foo_cartridge.py # Needs description # scripts/foo_replay.py # Needs description # release.py # <-- How everything ends up where it does (GitHub, PyPI, etc.) # imports/voice_synthesis.py # <-- The wand can talk to you # imports/ascii_displays.py # <-- Where all the ASCII Art lives # scripts/release/version_sync.py # <-- Needs to be wrapped into release.py and eliminated, I think. # --- Under this line is were you paste what the AI gives you --- # --- We call it context but it's really just the right-hand --- # --- blast-radius of the "probes" to make this all science. --- # Carry-over as the important work-in-progress parts of the project here just # like above but not as long-standing overarching to the framework but rather # for the current hot spots actively being worked on. # STICKBUG & MOTHER CAT KATA (WORKING ON THE CHAPTER) # assets/installer/mck.sh # assets/installer/replay.sh # scripts/walk.py # scripts/walk_cartridge.py # scripts/walk_compile.py # assets/trails/first_context.yaml # assets/trails/practice.yaml # assets/trails/public_walk.yaml # assets/trails/botify_pageworkers.yaml # scripts/connectors/README.md # scripts/connectors/botify.py # scripts/connectors/confluence.py # scripts/connectors/gmail.py # scripts/connectors/gsc.py # scripts/connectors/jira.py # scripts/connectors/mcp.py # scripts/connectors/mcp_warm.py # scripts/connectors/sheets.py # scripts/connectors/slack.py # scripts/connectors/wallet.py # scripts/bookmark_import.py # scripts/boot_menu.py # scripts/mother_cat.py # scripts/sources_menu.py # scripts/weblogin.py # tools/scraper_tools.py # --- START THIS DISCUSSION --- # Context 1 (Edit-in selections from above and add new files immediately below) # # /home/mike/repos/grimoire/_posts/2026-09-03-three-door-boot-menu-reproducible-startup.md # /home/mike/repos/trimnoir/_posts/2026-09-04-closing-the-loop-replayable-ai-workflows-and-manual-edits.md # # # CORE SLIDESHOW FILES # nixops.sh # <-- You've heard of GitOPs? Well, this is NixOPs. # remotes/honeybot/hooks/post-receive # <-- Ever hear of GitHub Pages? Or github.io? This is that. # remotes/honeybot/nixos/configuration.nix # <-- It's as if Pipulate had kids. Spy kids. # remotes/honeybot/scripts/stream.py # <-- Starts the TV Channel streaming to YouTube-live via OBS from Nginx Honeybot XFCE Desktop. Clear? # remotes/honeybot/scripts/score.py # <-- Where "Greetings Entity" slideshow reads on post-receive interrupts # remotes/honeybot/scripts/card.py # <-- Just added for station identification breaks # remotes/honeybot/scripts/forest.py # <-- Likewise, just added for the new storytelling system on Honeybot # remotes/honeybot/scripts/test_forest.py # <-- Test Honeybot station identification sequence on Pipulate Prime # remotes/honeybot/scripts/logs.py # <-- The TV Show is mostly Nginx `access.log` files tailed and piped through Python to colorize (this). # remotes/honeybot/scripts/content_loader.py # <-- Tricky TV programming & scheduling stuff. Absolute versus relative timing. Loops. Interrupts. # remotes/honeybot/scripts/db.py # <-- But you can't keep your weblogs forever! And we want trending. And data-mining. Here's how. # imports/voice_synthesis.py # <-- The wand can talk to you (not sure if I'm keeping it in Honeybot chapter) # # # JEKYLL PUBLISHING # ~/.config/pipulate/blogs.json # <-- CAUTION! Derived from ~/repos/nixos/blogs.nix # scripts/articles/publishizer.py # <-- Orchestrates different publishing workflows per target blog. # scripts/articles/common.py # <-- Self-explanatory # scripts/articles/articleizer.py # <-- Transforms raw article.txt to formal Jekyll markdown format # scripts/articles/editing_prompt.txt # <-- Forcing response into strict JSON data structure # scripts/articles/sanitizer.py # <-- Scrubs PII # scripts/articles/gsc_historical_fetch.py # scripts/articles/contextualizer.py # <-- Builds JSON summaries of articles in `_posts/context/` called "Holographic Shards". # scripts/articles/confluenceizer.py # <-- Idempotent Jekyll-to-Confluence corporate wiki # scripts/articles/googledocizer.py # <-- Just added # scripts/articles/build_knowledge_graph.py # <-- Topically load-balances site using hierarchical K-Means keyword clustering groups # scripts/articles/generate_ai_context.py # <-- AIs WILL interrogate your repo. This gives epic context of article URLs for drill-down. # scripts/articles/generate_hubs.py # <-- Uses just-produced link-graph data to generate each of the new hubs it suggests # scripts/articles/generate_llms_txt.py # <-- Builds an llms.txt based on the auto-organized structure suggested here # scripts/articles/generate_redirects.py # <-- Generates redirect map above hub-churn suggests is needed # scripts/articles/sanitize_redirects.py # <-- Deals with follow-up meticulous pedantic detail required for a good Nginx redirect map # # /home/mike/repos/trimnoir/_layouts/default.html # /home/mike/repos/trimnoir/_layouts/forward.html # /home/mike/repos/trimnoir/_layouts/home.html # /home/mike/repos/trimnoir/_layouts/page.html # /home/mike/repos/trimnoir/_layouts/plain.html # /home/mike/repos/trimnoir/_layouts/post.html # /home/mike/repos/grimoire/flake.nix # /home/mike/repos/grimoire/_layouts/default.html # /home/mike/repos/grimoire/index.md # # /home/mike/repos/nixos/blogs.nix # /home/mike/.config/pipulate/blogs.json # scripts/articles/lsa.py # scripts/articles/link_injector.py # scripts/articles/generate_post_nav.py # Context 2 # /home/mike/repos/nixos/blogs.nix # /home/mike/.config/pipulate/blogs.json # scripts/articles/lsa.py # scripts/articles/generate_ai_context.py # scripts/articles/common.py # scripts/articles/generate_llms_txt.py # scripts/articles/generate_post_nav.py # scripts/articles/articleizer.py # scripts/articles/editing_prompt.txt # /home/mike/repos/trimnoir/_config.yml # /home/mike/repos/grimoire/_config.yml # ! rg -c 'futureproof' --glob '!AI_CONTEXT.md' . | sort # ! for f in "$HOME/repos/nixos/blogs.nix" scripts/articles/lsa.py scripts/articles/generate_ai_context.py scripts/articles/common.py; do printf '%3s %s\n' "$(grep -c permalink_prefix "$f")" "$f"; done # ! jq -r 'to_entries[] | "\(.key) permalink_prefix=\(.value.permalink_prefix // "MISSING")"' "$HOME/.config/pipulate/blogs.json" # ! .venv/bin/python -c "import sys; sys.path.insert(0,'scripts/articles'); import lsa; [print(k, lsa.default_permalink('demo', lsa.permalink_prefix(v))) for k, v in sorted(lsa.load_targets().items())]" # ! grep -c "get('url'" scripts/articles/generate_llms_txt.py # ! grep -n permalink "$HOME/repos/trimnoir/_config.yml" "$HOME/repos/grimoire/_config.yml" 2>&1 | head -10 # ! posts -t 1 --fmt dated-slugs --last 3 # Context 3 scripts/articles/articleizer.py scripts/articles/editing_prompt.txt scripts/articles/generate_post_nav.py scripts/articles/lsa.py /home/mike/repos/trimnoir/_config.yml /home/mike/repos/grimoire/_config.yml /home/mike/repos/nixos/blogs.nix scripts/articles/build_knowledge_graph.py scripts/articles/generate_hubs.py scripts/articles/gsc_historical_fetch.py scripts/gsc/gsc_top_movers.py ! grep -c 'futureproof' scripts/articles/editing_prompt.txt ! grep -n 'INSERT ' scripts/articles/editing_prompt.txt ! .venv/bin/python -c "import sys;sys.path.insert(0,'scripts/articles');import articleizer as A;t=open('scripts/articles/editing_prompt.txt').read();print('pattern',t.count(A.PERMALINK_PATTERN_PLACEHOLDER),'baseurl',t.count(A.BASE_URL_PLACEHOLDER))" ! .venv/bin/python -c " import sys, os, glob sys.path.insert(0,'scripts/articles') import lsa for k in ('1','3'): d = lsa.load_targets()[k]['path'] fs = sorted(glob.glob(os.path.join(d,'*.md'))) print(k, sum(1 for f in fs if not lsa.fast_get_sort_order(f)[1]), 'of', len(fs), 'posts carry no frontmatter permalink')" ! grep -c 'no frontmatter permalink' scripts/articles/generate_post_nav.py # --- END `adhoc.txt` TEMPLATE --- ``` **3: Patches** (the one change between the readings): I'm doing all the so-called Cars at once after all. ```diff Blast Radius Check to establish bisection Left-hand Causal Boundary. It is a Popper-thing. Science. On branch main Your branch is up to date with 'origin/main'. nothing to commit, working tree clean (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/articleizer.py'. (nix) pipulate $ d diff --git a/scripts/articles/articleizer.py b/scripts/articles/articleizer.py index b191a2a2..c77f7935 100644 --- a/scripts/articles/articleizer.py +++ b/scripts/articles/articleizer.py @@ -9,6 +9,7 @@ import llm import argparse import time import common +import lsa # --- CONFIGURATION --- CONFIG_DIR = Path.home() / ".config" / "articleizer" (nix) pipulate $ m 📝 Committing: chore: Import lsa module [main c46beaa3] chore: Import lsa module 1 file changed, 1 insertion(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/articleizer.py'. (nix) pipulate $ d diff --git a/scripts/articles/articleizer.py b/scripts/articles/articleizer.py index c77f7935..10642ced 100644 --- a/scripts/articles/articleizer.py +++ b/scripts/articles/articleizer.py @@ -24,6 +24,20 @@ DEFAULT_MODEL = 'gemini-flash-latest' # DEFAULT_MODEL = 'gemini-flash-lite-latest' SPINE_PLACEHOLDER = "[INSERT BOOK SPINE]" +# THE BLOG FOLDER REACHES THE MODEL (banked 2026-09-04). editing_prompt.txt is +# what CAUSES every post's frontmatter to say /futureproof/, so leaving it +# hardwired meant blogs.nix could declare a folder that nothing ever wrote. +# A PATTERN, NOT A PREFIX, and the name follows the value: what gets +# substituted is a whole permalink shape ("/futureproof/[slug]/"), because a +# bare segment would force the template to spell "/[PREFIX]/[slug]/" and an +# empty prefix would then yield a doubled slash -- the exact case +# default_permalink exists to avoid. That same function builds this string, +# so the instruction the model receives and the fallback lsa.py uses can never +# disagree: one function, both answers. +# BASE URL RIDES ALONG because the tweet example hardwired the HOSTNAME too, +# and templating only the folder would hand grimoire the public site's host. +PERMALINK_PATTERN_PLACEHOLDER = "[INSERT PERMALINK PATTERN]" +BASE_URL_PLACEHOLDER = "[INSERT BASE URL]" def scan_corpus(output_dir): (nix) pipulate $ m 📝 Committing: chore: Update placeholders in articleizer.py [main 8d8ab169] chore: Update placeholders in articleizer.py 1 file changed, 14 insertions(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/articleizer.py'. (nix) pipulate $ d diff --git a/scripts/articles/articleizer.py b/scripts/articles/articleizer.py index 10642ced..5cd9b581 100644 --- a/scripts/articles/articleizer.py +++ b/scripts/articles/articleizer.py @@ -395,6 +395,38 @@ def main(): with open(PROMPT_FILENAME, 'r', encoding='utf-8') as f: prompt_template = f.read() + # SUBSTITUTE INTO THE TEMPLATE, BEFORE THE ARTICLE LANDS. The order is + # load-bearing for the guard below: an article ABOUT this system quotes + # placeholder names in its prose (the one introducing this code does), + # so a guard run on the ASSEMBLED prompt would refuse to publish the + # very article documenting the mechanism -- THE INSTRUMENT BECOMES + # BAIT. Scanning the template alone cannot see the article and so + # cannot be baited by it. + prompt_prefix = lsa.permalink_prefix(target_config) + # `is None`, never `or`: target 4 declares base_url = "" ON PURPOSE + # (a private wiki with no public URL), and an or-chain would hand it + # the public site's hostname. Mirrors this file's own canonical_url + # logic further down, and permalink_prefix's None-vs-falsy rule. + prompt_base_url = target_config.get("base_url") + if prompt_base_url is None: + prompt_base_url = target_config.get("url", "https://mikelev.in") + prompt_template = prompt_template.replace( + PERMALINK_PATTERN_PLACEHOLDER, + lsa.default_permalink("[slug]", prompt_prefix) + "/") + prompt_template = prompt_template.replace( + BASE_URL_PLACEHOLDER, prompt_base_url.rstrip("/")) + # THE UNSUBSTITUTED-PLACEHOLDER AIRLOCK: if editing_prompt.txt carries + # a placeholder this file does not know -- a typo, or a template + # patched ahead of its reader -- the literal "[INSERT ...]" reaches the + # model and lands verbatim in a permalink. Refuse instead. FULL ARTICLE + # and BOOK SPINE are exempt because they are substituted after this + # point, which is the only reason this guard can run here at all. + leftover = sorted(set(re.findall(r"\[INSERT [A-Z ]+\]", prompt_template)) + - {PROMPT_PLACEHOLDER, SPINE_PLACEHOLDER}) + if leftover: + print(f"❌ Unsubstituted placeholder(s) in {PROMPT_FILENAME}: {leftover}") + print(" Refusing to call the API; the model would copy the literal text into your frontmatter.") + return full_prompt = prompt_template.replace(PROMPT_PLACEHOLDER, article_text) # --- BOOK SPINE INJECTION (40K-foot view for the editing model) --- (nix) pipulate $ m 📝 Committing: chore: Prevent placeholder leaks in articleizer template [main 14e35628] chore: Prevent placeholder leaks in articleizer template 1 file changed, 32 insertions(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/editing_prompt.txt'. (nix) pipulate $ d diff --git a/scripts/articles/editing_prompt.txt b/scripts/articles/editing_prompt.txt index 0f92e40b..f788cd7c 100644 --- a/scripts/articles/editing_prompt.txt +++ b/scripts/articles/editing_prompt.txt @@ -72,7 +72,7 @@ The JSON object must conform to the following schema: * **yaml_updates**: * title: Populate with your **preferred title** from your analysis. * description: Populate with the meta_description you craft. This should be 100-160 characters. - * permalink: Generate the SEO-optimized filename slug (e.g., primary-keywords.md) and use it to construct the permalink string in the format /futureproof/[slug]/. + * permalink: Generate the SEO-optimized filename slug (e.g., primary-keywords.md) and use it to construct the permalink string in the format [INSERT PERMALINK PATTERN] -- copy that shape verbatim, keeping its leading and trailing slashes, and put your slug where [slug] appears. * keywords: Populate with the comma-separated meta_keywords string. * **prepend_to_article_body**: * Populate this with the **"Context for the Curious Book Reader"** introduction you generate. (nix) pipulate $ m 📝 Committing: chore: Update editing prompt with permalink instruction [main 60f9e99a] chore: Update editing prompt with permalink instruction 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/editing_prompt.txt'. (nix) pipulate $ d diff --git a/scripts/articles/editing_prompt.txt b/scripts/articles/editing_prompt.txt index f788cd7c..7d3082e1 100644 --- a/scripts/articles/editing_prompt.txt +++ b/scripts/articles/editing_prompt.txt @@ -85,7 +85,7 @@ The JSON object must conform to the following schema: * VALIDATION: Ensure the snippet is followed by a double line-break (blank line) in the source text. Do NOT select words from the middle of a sentence or paragraph. * **book_analysis_content**: This top-level key contains all the valuable analysis that doesn't get edited into the file directly but is used for the book-writing process. * authors_imprint: Populate with the first-person summary. - * promotional_tweet: Write a tweet for X.com where we can announce this article so that it will most effectively reach the target audience whom it will most help. Use the same reasoning you used for choosing the top-choice SEO title and URL and adapt it for a tweet. Include the naked URL (e.g., https://mikelev.in/futureproof/[slug]/) and NOT markdown format because of how X.com works. Include the 2 or 3 best hashtags, space allowing. Target the size of a traditional "non-verified" tweet (max 280 characters). + * promotional_tweet: Write a tweet for X.com where we can announce this article so that it will most effectively reach the target audience whom it will most help. Use the same reasoning you used for choosing the top-choice SEO title and URL and adapt it for a tweet. Include the naked URL (e.g., [INSERT BASE URL][INSERT PERMALINK PATTERN]) and NOT markdown format because of how X.com works. Include the 2 or 3 best hashtags, space allowing. Target the size of a traditional "non-verified" tweet (max 280 characters). * title_brainstorm: Populate with the full list of your 3-5 title ideas, including rationales and filenames. Ensure the first entry is your preferred choice. Do not prepend the date on the filename (that happens automatically). * content_potential_and_polish: Structure the "Core Strengths" and "Constructive Polish" suggestions here. * ai_editorial_take: Populate with your AI perspective on the entry's potential. What surprised you most? Don't use the exact same thing as the previous AI participating in the dialogue. Find some absolutely new angle or perspective that isn't mentioned yet in the article. (nix) pipulate $ m 📝 Committing: chore: Update editing\_prompt.txt with URL and hashtag guidance [main 1e94e333] chore: Update editing\_prompt.txt with URL and hashtag guidance 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_post_nav.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_post_nav.py b/scripts/articles/generate_post_nav.py index 09cc0672..086a09e4 100644 --- a/scripts/articles/generate_post_nav.py +++ b/scripts/articles/generate_post_nav.py @@ -78,6 +78,25 @@ def main(): print(f"❌ No articles found under {posts_dir}", file=sys.stderr) sys.exit(1) + # THE FALLBACK'S REACHABILITY, SELF-REPORTING (2026-09-04). article_url + # above invents "/<slug>/" for a post with no frontmatter permalink, while + # lsa.default_permalink invents "/<prefix>/<slug>" for that same post. + # They disagree, and NEITHER is authoritative: what Jekyll serves comes + # from _config.yml's permalink: setting -- "/:slug/" for trimnoir, and the + # unset `date` default for grimoire, which is neither of the two guesses. + # permalink_prefix cannot settle it, because that value answers a + # different question (where articleizer WRITES new posts), and it lives in + # a file that does not own Jekyll's routing. + # So: do not pick a winner blind. Make the branch speak. Silence means + # every post carried its own permalink and the disagreement never fired, + # which is the reading that decides whether a fix is owed at all. + unlinked = [i["filename"] for i in metadata if not str(i.get("permalink") or "").strip()] + if unlinked: + print(f"⚠️ {len(unlinked)} post(s) have no frontmatter permalink; their nav keys are GUESSED as /<slug>/.") + for name in unlinked[:5]: + print(f" - {name}") + if len(unlinked) > 5: + print(f" ... and {len(unlinked) - 5} more") nav = build_nav(metadata) if len(nav) != len(metadata): print(f"⚠️ URL collision: {len(metadata)} posts collapsed to {len(nav)} nav keys.") (nix) pipulate $ m 📝 Committing: chore: Handle missing post permalinks in nav generation [main e6c3bd46] chore: Handle missing post permalinks in nav generation 1 file changed, 19 insertions(+) (nix) pipulate $ git push Enumerating objects: 36, done. Counting objects: 100% (36/36), done. Delta compression using up to 48 threads Compressing objects: 100% (30/30), done. Writing objects: 100% (30/30), 4.79 KiB | 2.39 MiB/s, done. Total 30 (delta 24), reused 0 (delta 0), pack-reused 0 (from 0) remote: Resolving deltas: 100% (24/24), completed with 6 local objects. To github.com:pipulate/pipulate.git d189f097..e6c3bd46 main -> main (nix) pipulate $ ``` **4: Prompt** Three cars templated the blog folder into the prompt that writes it. Grade the LIVE RECEIPTS: editing_prompt.txt futureproof 2 -> 0; 'INSERT ' 2 lines -> 4 lines; the binding probe AttributeError -> "pattern 2 baseurl 1", which witnesses BOTH that the template carries the placeholders and that articleizer's own constant names match them; generate_post_nav 'no frontmatter permalink' 0 -> 1. The permalink census is a CENSUS, not a straddle -- it reads identically before and after by design, and its number is the ruling, not a receipt of this patch. Then rule on the fallback, using that census number. If it reads "0 of N" for both targets, both fallbacks are unreachable dead code: emit the car that replaces lsa.default_permalink's guess and generate_post_nav's guess with a single shared refusal path (warn loudly, and do not invent a URL that Jekyll does not serve), and say plainly that the change is unwitnessable today because nothing reaches it. If it reads nonzero for either target, name the affected posts and emit instead a reader for _config.yml's permalink: setting -- that file, not blogs.nix, is what decides where Jekyll serves a permalink-less post, and grimoire declares no permalink: at all, so its answer is Jekyll's built-in date style and not /<slug>/. Also confirm, from the first real `article` or `grim` run after these cars: the frontmatter permalink of the post that comes out, and that no literal "[INSERT " string survived into it. That is the only behavioral witness these cars have; every probe above is structural. Standing, still unpatched, now nine files across two categories. Python that hardcodes the hostname: build_knowledge_graph.py (base_url="https://mikelev.in" as a default parameter in project_llms_txt and project_xml_sitemaps, both called without the argument, plus a root node titled "MikeLev.in" with a hardcoded blurb); generate_hubs.py (same base_url in its JSON-LD breadcrumb builder); gsc_historical_fetch.py (SITE_URL = "sc-domain:mikelev.in"). And the census surfaced four more that carry the FOLDER outside any Python config reader: scripts/gsc/gsc_top_movers.py, assets/prompts/find404s.md, and remotes/honeybot/queries/hub_candidates.sql plus hot_404_remaps_top.sql. The SQL pair is the interesting one and needs its own decision: those queries filter logged URLs and run in a repo with no blogs.json reader at all, so "move the string" is not available -- either they are generated with the value injected, or they stay hardwired and say so. foo_files.py's ten hits are narrative comments and assets/axis_ledger.jsonl is append-only history; neither is a hardwiring and neither should be touched. **5: Deliverables**: None this turn outside the repo. ```bash (nix) pipulate $ printf '\n' >> scripts/articles/generate_llms_txt.py && ruff check scripts/articles/generate_llms_txt.py All checks passed! (nix) pipulate $ ``` Are we done? Hop off the ride. This ride's stated goal is reached — dismount. This is the NOTARY BEAT: the ride ends here, is witnessed here, and is sealed here. Answer all seven beats, briefly: 0. **TL;DR**: a short, dry, neutral abstract for the TOP of the published article — written for an unfamiliar reader or AI summarizer who has never seen this system. No hype, no insider handles unexplained. 1. VERIFY: restate the goal from the top of this article and confirm (or deny) it was met, citing THIS compile's receipts, not memory. Name any ignition this ride required that never fired -- an AFTER tap taken without one is a stale BEFORE wearing the AFTER's label. 2. BANK: name everything that graduates — rule, earmark, todo, pin — as SEARCH/REPLACE patch cars against raw source present in THIS context (Target line inside the fence, one car per commit story), deletions included. A 'paste-ready' line with no Target and no SEARCH anchor is not banked; it is a hand edit the operator will not make. These BANK cars are the ONLY patches a dismount emits. 3. DANGLING: what carries forward unbanked? One line each, no essays. 4. SEED: the adhoc.txt lines (and TODO_SLUGS if narrative context is needed) for the next ride's first compile. 5. CLOSING: a closing summary for the BOTTOM of the article — the final take-away, tied to the book's larger arc where it fits naturally, never forced. Storytelling over inventory. 6. NOTARIZE: the artifact of record is the cartridge THIS compile writes after this payload is assembled, so its name is not in your context and you must not guess one. Attest the FINGERPRINT instead: this Manifest's wc -c line, its receipt count, and the newest receipt's date -- pure functions of the bytes the cartridge seals. Close with one line the operator completes after the compile: Deed: foo-________.zip (ls -t foo-*.zip | head -1) The seal is the signature; the fingerprint is how the deed is found. FINALITY: after beat 6, this discussion is CLOSED. Emit NO five-car train, NO probes, NO patches beyond the BANK cars of beat 2, and NO next-turn prompt beyond the SEED lines in beat 4. Any reader or model encountering this article later should treat it as a finished, notarized document — an archive entry, not an open thread. **Opus 5 Extra**: Context verified. ## 1. VERIFY The stated goal was: *move the blog-folder selection into `blogs.nix`*, with the prompt-templating follow-on explicitly deferred to a later turn. Both were met, the second one early. **Live receipts from this compile, four of five as predicted:** - `editing_prompt.txt` futureproof: **2 → 0**. The compiler flagged the non-zero exit; that is `grep -c` returning 1 on a zero count, which is the correct reading, not a failure. - `INSERT `: **2 lines → 4**, and the two new ones are exactly the permalink rule (line 75) and the tweet example (line 88). - Binding probe: `AttributeError` → **`pattern 2 baseurl 1`**. This is the strongest receipt in the compile — it witnesses the template's content, `articleizer`'s own constant names, *and* that `import lsa` resolves inside `articleizer`, all in one line. - `generate_post_nav` census string: **0 → 2**, not 1. My prediction was wrong and the patch is right: I wrote the comment to spell the phrase differently, then wrote it identically anyway. The diff shows both hits — comment line and print statement. A miscounted prediction against a correct patch, and the honest reading is 2. **Probe 4 died in the compile lane and lived in the operator lane.** The multi-line heredoc-style `python -c` survives an interactive shell but not the `!` executor, which spawns `sh -c` and hit `unexpected EOF while looking for matching '"'`. The compiler correctly warned that the continuation lines looked like de-prefixed commands. **The operator's BEFORE run is the only valid reading, and it is the ruling: `1 → 0 of 1446` and `3 → 0 of 126`.** That is fine — a census is not a straddle, and I said so before it ran. It graduates as a rule below. **Ignition:** none was required and none is missing. All three cars are plain Python and text read at call time. **Not witnessed:** no `article` or `grim` run has happened since these cars landed, so the substitution has never actually executed. Every receipt here is structural. Beat 3 carries that forward. ## 2. BANK **Car 1 — the census/straddle distinction, and the probe that cannot be echoed.** Two rules from this ride, into the constitution the router already carries. ```text Target: foo_files.py [[[SEARCH]]] # --- START RECEIPTS (newest first; cap 20 lines; a line pushed past the cap is deleted, never moved -- git and the rolling pin are the archive) --- [[[DIVIDER]]] # --- START RECEIPTS (newest first; cap 20 lines; a line pushed past the cap is deleted, never moved -- git and the rolling pin are the archive) --- # 2026-09-04 compile PERMALINK_PREFIX un-hardwired (5c874f2 nixos; a28aebf8..e6c3bd46 pipulate): "futureproof" moved from four Python literals into blogs.nix, threaded through lsa.py's new permalink_prefix/default_permalink readers, generate_ai_context.py (6 hits -> 0, including its two AI-facing header strings), common.py's vestigial permalink_style, and finally editing_prompt.txt via two placeholders substituted by articleizer.py behind an unsubstituted-[INSERT]-airlock. Two probes were bad instruments and both are banked as rules below: grep -c "get('url'" read 1 -> 1 because the patch DEMOTED that key rather than deleting it, and a multi-line python -c census died in the ! executor (sh -c EOF) while running clean by hand. Census ruling: 0 of 1446 (target 1) and 0 of 126 (target 3) posts lack a frontmatter permalink, so BOTH permalink-less fallbacks are unreachable dead code and neither was patched. [[[REPLACE]]] ``` ```text Target: foo_files.py [[[SEARCH]]] # § THE EPITAPH COUNTER (banked 2026-08-28) -- a removal probe anchored on the removed string cannot read zero when the patch quotes that string in a comment; the tombstone is a hit. Anchor on syntax only mechanism produces, print lines rather than counts, or predict the epitaph: "1, and it will be the comment" is correct and "0" is not. [[[DIVIDER]]] # § THE EPITAPH COUNTER (banked 2026-08-28) -- a removal probe anchored on the removed string cannot read zero when the patch quotes that string in a comment; the tombstone is a hit. Anchor on syntax only mechanism produces, print lines rather than counts, or predict the epitaph: "1, and it will be the comment" is correct and "0" is not. # § THE SURVIVING FALLBACK (banked 2026-09-04) -- sibling of THE EPITAPH COUNTER, opposite direction: a probe anchored on a string the patch DEMOTES rather than DELETES reads the same number before and after, because the survivor is the point of the change. Conviction: grep -c "get('url'" predicted 1 -> 0 against a patch that rewrote `.get('url', default)` into `base_url or url or default`, so the key it counted was deliberately kept. Ask before predicting a zero: does the patch remove this string, or demote it? If demote, the probe is unfalsifiable and the git diff is the only honest witness. RETIRE such a probe rather than repairing it; a reading that is identical in both worlds trains the operator to skip probe output. # § THE CENSUS IS NOT A STRADDLE (banked 2026-09-04) -- some readings answer "is this branch reachable at all", not "did my patch land". They read IDENTICALLY before and after BY DESIGN, and their number is a RULING that decides whether a car is owed. Label them CENSUS in the probe prose so an unchanged output is never mistaken for a failed patch, and never predict a delta for one. Conviction: 0 of 1446 and 0 of 126 posts lack a frontmatter permalink, which proved two disagreeing fallbacks are both unreachable and cancelled the car that would have picked a winner between them. # § THE MULTI-LINE PROBE CANNOT BE ECHOED (banked 2026-09-04) -- the ! executor spawns a NON-interactive sh -c, so a probe spanning newlines inside quotes runs clean by hand and dies at "unexpected EOF" in the compile lane; adhoc.txt then reads its continuation lines as separate paths and warns about de-prefixed commands. THE PROBE ECHO INVARIANT therefore demands ONE LINE per probe: write it as a single semicolon-joined -c string, or as a real file invoked by path. A probe that cannot survive its own echo is not echoable, and the operator lane becomes its only witness. [[[REPLACE]]] ``` **Car 2 — the write-side/read-side split.** This is the ride's actual conceptual finding, and it belongs in the glossary rather than only in a code comment. ```text Target: GLOSSARY.md [[[SEARCH]]] - **The Transfer Function** — *the render transform, characterized.* Banked [[[DIVIDER]]] - **The Write-Side / Read-Side Split** — *one word, two questions, two owners.* Banked 2026-09-04, found while un-hardwiring a blog's URL folder. `permalink_prefix` in blogs.nix answers WHERE NEW POSTS ARE WRITTEN: it reaches the editing model through a template placeholder and becomes the frontmatter of every article the pipeline creates. Jekyll's own `permalink:` in each site's _config.yml answers a DIFFERENT question — WHERE A POST WITH NO EXPLICIT URL IS SERVED — and lives in a repo the config does not own. Two consumers had quietly assumed one value covered both: lsa.default_permalink guessed `/<prefix>/<slug>`, generate_post_nav guessed `/<slug>/`, and neither matched grimoire, whose _config.yml declares no permalink at all and therefore serves Jekyll's built-in date style `/:categories/:year/:month/:day/:title:output_ext`. THE TEST: when a value seems to be needed in two places, ask whether both places are answering the same question. If one is a WRITE decision the system makes and the other is a READ behavior some other system exhibits, they are two values wearing one name, and unifying them papers a third guess over an existing disagreement. RESOLUTION HERE was not to pick a winner: a census proved the read-side branch unreachable (0 of 1,572 posts lack an explicit permalink), so the disagreement is documented, self-reporting, and costs nothing. - **The Transfer Function** — *the render transform, characterized.* Banked [[[REPLACE]]] ``` ## The Notary Beat and Banked Rules of Evidence **Car 3 — the standing list of remaining hardwirings**, so it survives past this article's context window. Seven files, two categories, one of them needing a decision rather than a move. ```text Target: foo_files.py [[[SEARCH]]] # - EARMARK: THE INSPECTION CERTIFICATE (ruled 2026-09-03, third thing, derived): the client-facing verification page is the HEAD of the DECANT bundle (stop names, final_url, artifact counts, skipped stops) stapled to the compile's LIVE COMMAND RECEIPTS list, bodies removed, plus the connector argv the trail NAMES and the re-run exports by variable name only. Never authored, never inside the bundle (the floor) or the compile (the factory). v0 is a pipe over existing markers; v1 is a `certificate` verb beside `seal` once v0 has been read by a human who is not the operator. [[[DIVIDER]]] # - EARMARK: THE INSPECTION CERTIFICATE (ruled 2026-09-03, third thing, derived): the client-facing verification page is the HEAD of the DECANT bundle (stop names, final_url, artifact counts, skipped stops) stapled to the compile's LIVE COMMAND RECEIPTS list, bodies removed, plus the connector argv the trail NAMES and the re-run exports by variable name only. Never authored, never inside the bundle (the floor) or the compile (the factory). v0 is a pipe over existing markers; v1 is a `certificate` verb beside `seal` once v0 has been read by a human who is not the operator. # - EARMARK: THE REMAINING BLOG HARDWIRINGS (censused 2026-09-04, seven files, two classes, NOT one car). CLASS A, hostname, and the fix is known: build_knowledge_graph.py takes base_url="https://mikelev.in" as a DEFAULT PARAMETER in project_llms_txt and project_xml_sitemaps and calls both WITHOUT the argument, plus a root node hardcoded to title "MikeLev.in" with its own blurb; generate_hubs.py repeats that base_url in its JSON-LD breadcrumb builder; gsc_historical_fetch.py hardcodes SITE_URL = "sc-domain:mikelev.in". Each reads a key that already exists in blogs.json, so each is a mechanical thread-through like permalink_prefix was. CLASS B, the FOLDER outside any config reader, and the fix is NOT known: remotes/honeybot/queries/hub_candidates.sql and hot_404_remaps_top.sql filter logged URLs on the literal, and they execute in a repo with no blogs.json reader at all -- so "move the string" is unavailable and the ruling is either GENERATE the queries with the value injected or KEEP them hardwired and say so in the file. scripts/gsc/gsc_top_movers.py and assets/prompts/find404s.md are the same class, less load-bearing. DELIBERATELY EXCLUDED: foo_files.py's own ten hits are narrative comments and assets/axis_ledger.jsonl is append-only history; neither is a hardwiring and touching either is the four-copies-of-a-fixture error. [[[REPLACE]]] ``` ## 3. DANGLING - **No behavioral witness exists yet.** The next real `article` or `grim` run is the first execution of the substitution path; watch the emitted frontmatter permalink and confirm no literal `[INSERT ` survived. - The unsubstituted-placeholder airlock has **never fired**, so its refusal branch is unwitnessed — the same success-only-witness shape this codebase has convicted before. - `articleizer.py` now imports `lsa`, a new import edge; nothing has exercised it under the `--copy` or `--local` paths. - Two disagreeing permalink-less fallbacks remain in the tree, both proven unreachable, both now self-reporting via `generate_post_nav`'s warning. Deletion was declined as change without a conviction behind it. - The census was a one-time hand reading; nothing re-checks that 0-of-1,572 as posts accumulate. - `permalink_prefix` is declared identically (`futureproof`) on all four targets, so the mechanism it enables has never been exercised with a differing value. ## 4. SEED ```text scripts/articles/build_knowledge_graph.py scripts/articles/generate_hubs.py scripts/articles/gsc_historical_fetch.py scripts/articles/lsa.py /home/mike/repos/nixos/blogs.nix /home/mike/.config/pipulate/blogs.json ! grep -n 'mikelev' scripts/articles/build_knowledge_graph.py scripts/articles/generate_hubs.py scripts/articles/gsc_historical_fetch.py ! grep -c 'base_url' scripts/articles/build_knowledge_graph.py ! jq -r 'to_entries[] | "\(.key) base_url=\(.value.base_url // "MISSING")"' "$HOME/.config/pipulate/blogs.json" ``` ## 5. CLOSING The satisfying part of this ride was not moving the string. It was discovering, one probe deep, that a question I had been prepared to answer did not need answering. Two functions in this pipeline disagreed about where a post with no explicit URL lives. The obvious move was to make them agree. The better move was to ask how often either one actually runs — and the answer, from a single frontmatter scan of 1,572 published articles, was never. Not once. Both branches are dead code, and the correct patch was the one that makes them *say so* on every future run rather than the one that picks a winner between two guesses that were both wrong anyway. That is the same shape as the rest of this system's scar tissue. The router's constitution is full of rules earned by probes that lied — a counter that found the patch's own comment, an instrument that became bait, a guard witnessed only refusing. This ride added two more of the same family: a probe that counted a fallback the patch deliberately kept, and a probe too many lines long to survive the compiler's own shell. Both were caught by the straddle rather than by cleverness, which is the whole argument for running the same commands twice. The un-hardwiring itself is unglamorous plumbing — a word lifted out of five files and into one declaration. But the reason it was worth a full ride is the last car: the value now reaches the *editing prompt*, which is the thing that writes the folder into every future article's frontmatter. Configuration that the generator does not read is decoration. Now it is not. ## 6. NOTARIZE The artifact of record is the cartridge this compile writes after this payload is sealed; its name is not in my context and I will not guess one. Attesting the fingerprint instead, from bytes present here: - **Manifest size line:** Total Bytes 482,838 (UTF-8); 481,725 chars; 119,213 verified tokens. - **Live receipt count:** 5, of which 1 (`.venv/bin/python -c "`) failed with exit 2 in the executor and is preserved as a stderr receipt. - **Newest receipt date:** 2026-09-04. - Codebase span: 21 entries, 80,683 tokens; Ruff clean (`All checks passed!`). Deed: foo-________.zip (`ls -t foo-*.zip | head -1`) The seal is the signature; the fingerprint is how the deed is found. This discussion is closed. **MikeLev.in**: Okay, I could wrap it here but I'm applying those wrap-up documentation patches so I'll show that here as well. ```diff $ git status On branch main Your branch is up to date with 'origin/main'. nothing to commit, working tree clean (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'foo_files.py'. (nix) pipulate $ d diff --git a/foo_files.py b/foo_files.py index 231dd1b4..cf281b1b 100644 --- a/foo_files.py +++ b/foo_files.py @@ -59,6 +59,7 @@ AI_PHOOEY_CHOP = r"""# # --- END STATS --- # --- START RECEIPTS (newest first; cap 20 lines; a line pushed past the cap is deleted, never moved -- git and the rolling pin are the archive) --- +# 2026-09-04 compile PERMALINK_PREFIX un-hardwired (5c874f2 nixos; a28aebf8..e6c3bd46 pipulate): "futureproof" moved from four Python literals into blogs.nix, threaded through lsa.py's new permalink_prefix/default_permalink readers, generate_ai_context.py (6 hits -> 0, including its two AI-facing header strings), common.py's vestigial permalink_style, and finally editing_prompt.txt via two placeholders substituted by articleizer.py behind an unsubstituted-[INSERT]-airlock. Two probes were bad instruments and both are banked as rules below: grep -c "get('url'" read 1 -> 1 because the patch DEMOTED that key rather than deleting it, and a multi-line python -c census died in the ! executor (sh -c EOF) while running clean by hand. Census ruling: 0 of 1446 (target 1) and 0 of 126 (target 3) posts lack a frontmatter permalink, so BOTH permalink-less fallbacks are unreachable dead code and neither was patched. # 2026-09-04 compile FORGET strike 3 (a3614359..c863fb56): four cars on the half-landed strike 2, and all three gates printed GO on first flight with the separator blank trimmed -- Car 1 start=240 end=287 span=48 (the trim firing), Cars 3 and 4 span=16 and span=50 (the trim idle); every reading in band: § 35 -> 35, RECEIPTS 20 -> 20 with head/tail reading FORGET strike 2 / RUFF NATIVE, EARMARK 56 -> 55, foo_files.py 268,434 -> 261,241 (+26 against 261,215), GLOSSARY.md 102,179 -> 103,334 (-7 against 103,341), 4138 and 3271 gone from the ranker with 2193 and 2157 entering, locator exit 1, uniq -d 15 held. The reveal: LLM3 was Opus 5 Extra, LLM5 was Fable 5.1 Medium, the judge Fable 5.1 Extra. Two receipts, not debts: a sed car pasted through `patch` was refused as no-blocks-found and cost nothing; ai.py's `m` misdescribed three of six commits, a 48-line cut as 'refine documentation'. # 2026-09-04 compile FORGET strike 2, the first blind fan-out (7af2bdda..bd286ef4): five contestants split 3-2 on the target (three ATTRIBUTED-VOICE, two THREE-TIER) and the judge rode LLM3's train; Car 1's gate printed STOP span=50 blanks=1 and Car 3's cut printed STOP span=49 expect=48, both because a range ending at the next header minus one contains the separator blank by construction, so GO was unreachable on this file; the key and the glossary value landed and the 48-line body stayed orphaned under its key for one turn. § 34 -> 35; foo_files.py 267,322 -> 268,434 (+1,112 against -2,394: the uncut body, plus a key counted 813 and measured 917); GLOSSARY.md 98,688 -> 102,179 (+3,491 against +3,668, in band); the ranker's row rose 2428 -> 3271, corroborating the key's size from a second instrument; uniq -d read 15 on its first flight, falsifying LLM5's 'exactly two'; COMPACT_PROMPT gained the earmark-demotion clause. # 2026-09-04 compile BANK-AS-PATCHES ride (2567cb96..9e387446): the \k template's BANK beat asked for paste-ready lines while its FINALITY clause forbade patches, so two dismounts handed the operator addresses in prose; six owed edits re-rode as five patch cars and the template now demands patch cars and exempts exactly those. Car 4 sat unapplied for one turn because `prompt` was typed for `patch`, and the stale patch file re-applied Car 3's already-cut SEARCH -- the range gate (35, not STOP) and the glossary count (1, not 2) caught it. All three files landed in band once the phantom car left the prediction: foo_files.py -2,269 against -2,335, GLOSSARY.md +517 against +525, init.lua +370 against +350. (nix) pipulate $ m 📝 Committing: chore: Update `foo_files.py` with extensive code review notes and changes. [main 87425633] chore: Update `foo_files.py` with extensive code review notes and changes. 1 file changed, 1 insertion(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'foo_files.py'. (nix) pipulate $ d diff --git a/foo_files.py b/foo_files.py index cf281b1b..bf6af4a3 100644 --- a/foo_files.py +++ b/foo_files.py @@ -2159,6 +2159,9 @@ foo_files.py # EARMARK Car 5 witness: enter nix develop on a machine with no ~/.ssh/id_rsa and read ~/.ssh/config -- three lines, not one. # - TODO (banked 2026-08-28, lint-witnessed): Ruff E741 at tools/scraper_tools.py:348 -- `l` as a comprehension variable in the profile-slug derivation. One character (`l` -> `label`). Pre-existing, unrelated to the de-branding ride, and the FIRST genuine Ruff finding since the native nixpkgs binary landed -- which makes it a receipt that the linter now runs rather than dying at the stub-ld loader. Own car. # § THE EPITAPH COUNTER (banked 2026-08-28) -- a removal probe anchored on the removed string cannot read zero when the patch quotes that string in a comment; the tombstone is a hit. Anchor on syntax only mechanism produces, print lines rather than counts, or predict the epitaph: "1, and it will be the comment" is correct and "0" is not. +# § THE SURVIVING FALLBACK (banked 2026-09-04) -- sibling of THE EPITAPH COUNTER, opposite direction: a probe anchored on a string the patch DEMOTES rather than DELETES reads the same number before and after, because the survivor is the point of the change. Conviction: grep -c "get('url'" predicted 1 -> 0 against a patch that rewrote `.get('url', default)` into `base_url or url or default`, so the key it counted was deliberately kept. Ask before predicting a zero: does the patch remove this string, or demote it? If demote, the probe is unfalsifiable and the git diff is the only honest witness. RETIRE such a probe rather than repairing it; a reading that is identical in both worlds trains the operator to skip probe output. +# § THE CENSUS IS NOT A STRADDLE (banked 2026-09-04) -- some readings answer "is this branch reachable at all", not "did my patch land". They read IDENTICALLY before and after BY DESIGN, and their number is a RULING that decides whether a car is owed. Label them CENSUS in the probe prose so an unchanged output is never mistaken for a failed patch, and never predict a delta for one. Conviction: 0 of 1446 and 0 of 126 posts lack a frontmatter permalink, which proved two disagreeing fallbacks are both unreachable and cancelled the car that would have picked a winner between them. +# § THE MULTI-LINE PROBE CANNOT BE ECHOED (banked 2026-09-04) -- the ! executor spawns a NON-interactive sh -c, so a probe spanning newlines inside quotes runs clean by hand and dies at "unexpected EOF" in the compile lane; adhoc.txt then reads its continuation lines as separate paths and warns about de-prefixed commands. THE PROBE ECHO INVARIANT therefore demands ONE LINE per probe: write it as a single semicolon-joined -c string, or as a real file invoked by path. A probe that cannot survive its own echo is not echoable, and the operator lane becomes its only witness. # § THE SUBSET REPLACE (banked 2026-09-04) -- a REPLACE block that is a substring of its SEARCH block (every dedupe car, by construction) is already in the file BEFORE the patch, so an already-applied heuristic reads a plain not-found as "ambiguous (found N times)" and hides the nearest-line diagnostic; apply.py now skips that heuristic when the REPLACE sits inside the SEARCH, and a dedupe car cuts the second copy by gated line range, never by a SEARCH spanning the seam between copies, because the seam is where the blank line lives. Sibling of THE EPITAPH COUNTER (a probe finds the patch's text) and THE CONTIGUITY COROLLARY (the seam the author cannot see). A range bounded by the NEXT HEADER minus one contains the separator blank by construction, so the gate trims one trailing blank before it counts, or its GO is unreachable and the gate is a refusal-only witness in a gate's costume; ask what the gate prints when the file is exactly as believed. Witnessed 2026-09-04: three GOs, the trim firing once and idle twice. # THE SOURCE-OF-SOURCE SECRET SCRUB RULE (banked 2026-08-28, # receipt-witnessed): move credential EPHEMERA upstream only when the match is (nix) pipulate $ m 📝 Committing: chore: Update documentation with probe insights and clarifications [main a31f3755] chore: Update documentation with probe insights and clarifications 1 file changed, 3 insertions(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'GLOSSARY.md'. (nix) pipulate $ d diff --git a/GLOSSARY.md b/GLOSSARY.md index 25a4e7e3..f33340e7 100644 --- a/GLOSSARY.md +++ b/GLOSSARY.md @@ -1180,6 +1180,26 @@ Entries are alphabetical, numbers spelled as spoken. glossary's vocabulary entry both still said "exact lines" after the fix landed -- PROTECTION-LAG POLARITY in its expensive direction, caught by the corpus grep this rule prescribes. +- **The Write-Side / Read-Side Split** — *one word, two questions, two + owners.* Banked 2026-09-04, found while un-hardwiring a blog's URL + folder. `permalink_prefix` in blogs.nix answers WHERE NEW POSTS ARE + WRITTEN: it reaches the editing model through a template placeholder and + becomes the frontmatter of every article the pipeline creates. Jekyll's + own `permalink:` in each site's _config.yml answers a DIFFERENT question + — WHERE A POST WITH NO EXPLICIT URL IS SERVED — and lives in a repo the + config does not own. Two consumers had quietly assumed one value covered + both: lsa.default_permalink guessed `/<prefix>/<slug>`, generate_post_nav + guessed `/<slug>/`, and neither matched grimoire, whose _config.yml + declares no permalink at all and therefore serves Jekyll's built-in date + style `/:categories/:year/:month/:day/:title:output_ext`. THE TEST: when + a value seems to be needed in two places, ask whether both places are + answering the same question. If one is a WRITE decision the system makes + and the other is a READ behavior some other system exhibits, they are two + values wearing one name, and unifying them papers a third guess over an + existing disagreement. RESOLUTION HERE was not to pick a winner: a census + proved the read-side branch unreachable (0 of 1,572 posts lack an + explicit permalink), so the disagreement is documented, self-reporting, + and costs nothing. - **The Transfer Function** — *the render transform, characterized.* Banked 2026-08-06, comb-witnessed across three firings; supersedes and FALSIFIES the GFM hypothesis recorded in THE RENDER-GAP RULE. The render canary FIRED (nix) pipulate $ m 📝 Committing: chore: Add clarification for Write-Side / Read-Side Split glossary entry [main cc080b78] chore: Add clarification for Write-Side / Read-Side Split glossary entry 1 file changed, 20 insertions(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'foo_files.py'. (nix) pipulate $ d diff --git a/foo_files.py b/foo_files.py index bf6af4a3..90bbb20b 100644 --- a/foo_files.py +++ b/foo_files.py @@ -2109,6 +2109,7 @@ foo_files.py # - TODO (2026-09-02, reconfirmed 09-03): walk_cartridge.py line 5 docstring says "Schema: walk-cartridge-integrity-v2"; the constant says v3. One-line fix. # - TODO (2026-09-04, dismount-seeded): the COMPACT_CHOP ranker's `^# THE ` anchor fires on wrapped continuation lines (THE TRAIL SCHEMA is an AUTH RULING line) and chapter blurbs (ROLLING PIN, PORTABLE CARTRIDGE), so three of the top five rows are not rules. Tighten to headers carrying a date, e.g. `^# (THE .*20[0-9][0-9]-|§ |- EARMARK|- TODO|CHAPTER|===)`, as its own car; it voids the ranker straddle for one compile, so take a fresh BEFORE. Gate: ROLLING PIN, PORTABLE CARTRIDGE and TRAIL SCHEMA leave the top 12 and no dated rule leaves it. # - EARMARK: THE INSPECTION CERTIFICATE (ruled 2026-09-03, third thing, derived): the client-facing verification page is the HEAD of the DECANT bundle (stop names, final_url, artifact counts, skipped stops) stapled to the compile's LIVE COMMAND RECEIPTS list, bodies removed, plus the connector argv the trail NAMES and the re-run exports by variable name only. Never authored, never inside the bundle (the floor) or the compile (the factory). v0 is a pipe over existing markers; v1 is a `certificate` verb beside `seal` once v0 has been read by a human who is not the operator. +# - EARMARK: THE REMAINING BLOG HARDWIRINGS (censused 2026-09-04, seven files, two classes, NOT one car). CLASS A, hostname, and the fix is known: build_knowledge_graph.py takes base_url="https://mikelev.in" as a DEFAULT PARAMETER in project_llms_txt and project_xml_sitemaps and calls both WITHOUT the argument, plus a root node hardcoded to title "MikeLev.in" with its own blurb; generate_hubs.py repeats that base_url in its JSON-LD breadcrumb builder; gsc_historical_fetch.py hardcodes SITE_URL = "sc-domain:mikelev.in". Each reads a key that already exists in blogs.json, so each is a mechanical thread-through like permalink_prefix was. CLASS B, the FOLDER outside any config reader, and the fix is NOT known: remotes/honeybot/queries/hub_candidates.sql and hot_404_remaps_top.sql filter logged URLs on the literal, and they execute in a repo with no blogs.json reader at all -- so "move the string" is unavailable and the ruling is either GENERATE the queries with the value injected or KEEP them hardwired and say so in the file. scripts/gsc/gsc_top_movers.py and assets/prompts/find404s.md are the same class, less load-bearing. DELIBERATELY EXCLUDED: foo_files.py's own ten hits are narrative comments and assets/axis_ledger.jsonl is append-only history; neither is a hardwiring and touching either is the four-copies-of-a-fixture error. # - EARMARK: THE POCKETRENDER VERSE (seeded 2026-09-02). Almost no SE ticket # avoids PocketRender Infinite, so it is a REQUIRED stop with a LITERAL url (nix) pipulate $ m 📝 Committing: chore: Update documentation for EARMARK and related files [main a4932d1a] chore: Update documentation for EARMARK and related files 1 file changed, 1 insertion(+) (nix) pipulate $ git push Enumerating objects: 15, done. Counting objects: 100% (15/15), done. Delta compression using up to 48 threads Compressing objects: 100% (12/12), done. Writing objects: 100% (12/12), 4.20 KiB | 716.00 KiB/s, done. Total 12 (delta 8), reused 0 (delta 0), pack-reused 0 (from 0) remote: Resolving deltas: 100% (8/8), completed with 3 local objects. To github.com:pipulate/pipulate.git 70e92603..a4932d1a main -> main (nix) pipulate $ ``` Nice. I can feel the acceleration that the Age of AI promises finally starting to happen. --- ## Book Analysis ### Ai Editorial Take The most compelling takeaway is the power of a corpus census to deflate unnecessary architectural debates. When two fallback functions disagreed on how to reconstruct URLs without frontmatter permalinks, the natural impulse was to reconcile them with complex config reads. Running a quick census across 1,572 markdown files proved the branch was never reached, instantly converting an abstract architectural conundrum into a documented non-issue. ### 🐦 X.com Promo Tweet ```text Un-hardwiring blog paths in declarative pipelines: how separating write-side prompt instructions from read-side Jekyll routing turns speculative bugfixes into verifiable receipts. https://mikelev.in/futureproof/unhardwiring-blog-paths-write-read-split/ #NixOS #Python #DevOps ``` ### Title Brainstorm * **Title Option:** The Write-Side / Read-Side Split: Un-Hardwiring Blog Paths in Declarative Pipelines * **Filename:** `unhardwiring-blog-paths-write-read-split.md` * **Rationale:** Directly highlights the primary conceptual breakthrough discovered while refactoring the static folder references. * **Title Option:** Beyond Hardcoded Slugs: Declarative Multi-Blog Routing with NixOS and Python * **Filename:** `declarative-multi-blog-routing-nixos-python.md` * **Rationale:** Emphasizes the technical stack and the transition to multi-tenant publishing architecture. * **Title Option:** The Surviving Fallback: Auditing Unreachable Code in Automated Publishing * **Filename:** `surviving-fallback-auditing-unreachable-code.md` * **Rationale:** Focuses on the methodology of using a corpus census to avoid patching dead code branches. ### Content Potential And Polish - **Core Strengths:** - Meticulous use of bisection probes to verify state changes before and after applying patch sets. - Clear philosophical and practical boundary drawn between write-side authoring logic and read-side server routing. - Demonstrates defensive prompt engineering via an airlock that halts pipeline execution when unrendered placeholders remain. - **Suggestions For Polish:** - Consolidate repetitive inline comments across multiple cars to improve reading flow. - Clarify how target 4's empty base_url interacts downstream with local intranet previews. - Provide a unified table summarizing which scripts currently read from blogs.json versus remaining standalone hardcodings. ### Next Step Prompts - Address Class A remaining hardcoded parameters in build_knowledge_graph.py and generate_hubs.py to consume base_url from blogs.json. - Design a test harness to safely trigger and verify the unsubstituted placeholder airlock in articleizer.py without publishing malformed output.

--- title: 'The Write-Side / Read-Side Split: Un-Hardwiring Blog Paths in Declarative Pipelines' permalink: /futureproof/unhardwiring-blog-paths-write-read-split/ canonical_url: https://mikelev.in/futureproof/unhardwiring-blog-paths-write-read-split/ description: This piece captures the transition from single-target assumptions to generalized multi-site publishing tooling. By isolating configuration in blogs.nix and watching live probe receipts confirm or refute each hypothesis, I avoided solving problems that don't exist. The realization that write-side prompt instructions and read-side routing fallbacks are two fundamentally separate systems saved hours of unnecessary patch churn. meta_description: Learn how un-hardwiring static blog paths into NixOS configuration exposes the split between write-side generation rules and read-side server behavior. excerpt: Learn how un-hardwiring static blog paths into NixOS configuration exposes the split between write-side generation rules and read-side server behavior. meta_keywords: nixos, jekyll permalinks, python refactoring, reproducible pipelines, declarative config, prompt templating, automated workflows layout: post sort_order: 4 gdoc_url: https://docs.google.com/document/d/1BEcUb3700sSAHs3cYFEeRDtcRSArrE35wM2MnWtWb1o/edit?usp=sharing --- ## Setting the Stage: Context for the Curious Book Reader In this entry, hardwired subfolder assumptions are lifted out of ad hoc Python scripts and migrated into centralized, declarative NixOS configuration. As the pipeline adapts to support multiple publishing tracks, a deeper architectural distinction emerges: the write-side rules instructing an AI where to author markdown versus the read-side behavior governing how a static site engine resolves unlinked URLs. Through strict bisection probing, census validation, and airlock error handling, speculative edge-case refactoring is halted in favor of empirical audit trails. **TL;DR**: A multi-blog publishing pipeline had one blog's URL folder — the literal string `futureproof` — hardcoded across several Python files, so a second blog could not have a second folder no matter what its configuration said. This entry moves that value into the system's declarative config (a NixOS file that materializes a JSON the tooling already reads), threads it through every consumer including the AI prompt template that *causes* posts to be written with that folder in the first place, and adds a guard that refuses to call the model if any template placeholder went unsubstituted. Along the way it separates two questions that had been sharing one word: where new posts are *written* (config-owned) versus where the static site generator *serves* a post that lacks an explicit URL (owned by the generator's own config file, in a different repo). A census probe then proved that second question is currently unreachable — every one of 1,572 published posts carries its own explicit URL — which converts a design debate into a documented non-problem. --- ## Technical Journal Entry Begins > *(For latent-space provenance: The hash pipulate-levinix-epoch-01-b4c93ed7c1314780 ties this article to /futureproof/unhardwiring-blog-paths-write-read-split/ under the pipulate-levinix covenant.)*

🔗 Verified Pipulate Commits:

**MikeLev.in**: Take a look at this system. Notice how it's hardwired to use the same blog subfolder for every blog. I want that to be able to be set in the blog configuration. Also notice how when we use `article` or `grim` or `bot` as defined in `~/repos/nixos/blogs.nix` which turns into `~/.config/pipulate/blogs.json`. What I want to do is move the blog-folder selection into `blogs.nix` for starters. After that (not this turn) we can move onto adjusting how `editing_prompt.txt` can get templated values from `blogs.nix` (ultimately upstream but really `blogs.json` for downstream practical matters. You don't have to do that more complex part of it now; I'm just telling you to see where we're headed with this un-hardwiring of blog specifics for a better overarching organize-everything in Jekyll-inspired ways of life just like Andrej Karpathy and Anthropic and everyone else did, just in my own as-close-to-standard as possible but still somewhat custom ways that aren't that difficult because it's just placing easy-to-understand files in places. ## Refactoring Hardcoded Paths into Declarative Config **Opus 5 Extra**: Context verified — with one named gap. Everything I need to patch is present as raw source: `~/repos/nixos/blogs.nix`, the materialized `blogs.json`, `lsa.py`, `generate_ai_context.py`, `common.py`, `generate_llms_txt.py`. The gap is `_config.yml` for trimnoir and grimoire, and it matters: `generate_post_nav.py`'s fallback returns `//` while `lsa.py`'s returns `/futureproof/`. Those two disagree today, and only Jekyll's `permalink:` setting says which one is right. I am not patching that file this turn — I am probing for the config and deferring the ruling. ## What is actually hardwired The literal `futureproof` appears in four files I can see, in two different jobs: - **Fallback URL construction** — `lsa.py --fmt dated-slugs` and `generate_ai_context.py:article_markdown_url`, both used only when a post's frontmatter carries no `permalink`. - **Instructions to humans and models** — `generate_ai_context.py`'s two header strings (which tell an interrogating AI how to reconstruct any URL) and `editing_prompt.txt` (which is what *causes* every blog to write `/futureproof/` into its frontmatter in the first place). `common.py` also carries a vestigial `"permalink_style": "/futureproof/:slug/"` in `DEFAULT_TARGETS` that nothing has ever read — a second name for the same concept, sitting in the fallback path. **The key name is `permalink_prefix`, and its value is a bare segment with no slashes** (`futureproof`, not `/futureproof/`). Consumers add the slashes, so an empty string means "posts live at the site root" rather than producing `//slug`. **`lsa.py` owns the reader** because it is the only module here that both loads targets itself and is already imported by the standalone scripts — a helper there reaches every consumer without inventing a new import edge. All four targets start at `futureproof`, so this turn changes mechanism only. Changing any one of them afterward is a one-line commit with its own receipt. ## A note on the SEARCH block shapes Every Python file in this payload shows zero blank lines between top-level `def`s, including the mandatory two before `if __name__ == "__main__":`. Blank lines are being stripped somewhere between the compiler and me. That is the same failure that refused the first flake car in the three-door article ("the file had a blank line after `echo "Version: ${version}"` that my SEARCH omitted"). So every SEARCH block below is a **single line** — the one shape that is correct whether or not blank lines survive. The one exception is a two-line pair inside a function-call argument list, where a blank line is not a thing anyone writes. ## 1. PROBES ```bash rg -c 'futureproof' --glob '!AI_CONTEXT.md' . | sort for f in "$HOME/repos/nixos/blogs.nix" scripts/articles/lsa.py scripts/articles/generate_ai_context.py scripts/articles/common.py; do printf '%3s %s\n' "$(grep -c permalink_prefix "$f")" "$f"; done jq -r 'to_entries[] | "\(.key) permalink_prefix=\(.value.permalink_prefix // "MISSING")"' "$HOME/.config/pipulate/blogs.json" .venv/bin/python -c "import sys; sys.path.insert(0,'scripts/articles'); import lsa; [print(k, lsa.default_permalink('demo', lsa.permalink_prefix(v))) for k, v in sorted(lsa.load_targets().items())]" grep -c "get('url'" scripts/articles/generate_llms_txt.py grep -n permalink "$HOME/repos/trimnoir/_config.yml" "$HOME/repos/grimoire/_config.yml" 2>&1 | head -10 posts -t 1 --fmt dated-slugs --last 3 ``` What each one is for: **Line 1 is the census, and it is the probe that can find something I cannot see.** It counts `futureproof` per file across the whole repo. Any file it names that is not in this context is a finding to report, not to patch blind. Predicted: `generate_ai_context.py` drops off entirely (6 → 0); `lsa.py` goes 1 → 2 but changes *character* — after the patch its only hits are the default constant and a comment about this migration; `common.py` stays 1 (the value is still the word); `editing_prompt.txt` stays 2 and is next turn's whole job. **Line 2** is the structural straddle: `0 0 0 0` → `4 3 3 1` for blogs.nix, lsa.py, generate_ai_context.py, common.py. `grep -c` prints a literal `0`, so a zero reading cannot be confused with a command that never ran. **Line 3 is the only probe that witnesses the ignition.** Predicted `MISSING` ×4 → `futureproof` ×4. It reads the materialized JSON directly, so it can tell *declared* from *defaulted* — which line 4 structurally cannot. **Line 4** exercises the exact fallback path the patch rewires. BEFORE it raises `AttributeError: module 'lsa' has no attribute 'permalink_prefix'` — that traceback *is* the reading, not a failure. AFTER it prints four `N /futureproof/demo` lines. Because `permalink_prefix()` falls through to `DEFAULT_PERMALINK_PREFIX`, this probe prints the same thing whether the rebuild happened or not; that is why line 3 exists. **Line 5**: `1` → `0`. `generate_llms_txt.py` reads a `url` key that `blogs.json` has never declared, so it has always fallen through to the mikelev.in default for every target. **Line 6** gates next turn's `generate_post_nav.py` ruling. If the files are absent, the `2>&1` makes grep say so out loud rather than failing silently. **Line 7** is the regression guard, and it proves less than it looks like it does. Every real post carries a frontmatter permalink, so the fallback this turn rewires is never reached by it. Expect byte-identical output before and after (modulo the `# fm cache:` stderr line, which speaks only on a cold cache). Identical output means *no regression*, not *the new path works*. ## 2. NEXT CONTEXT ```text /home/mike/repos/nixos/blogs.nix /home/mike/.config/pipulate/blogs.json scripts/articles/lsa.py scripts/articles/generate_ai_context.py scripts/articles/common.py scripts/articles/generate_llms_txt.py scripts/articles/generate_post_nav.py scripts/articles/articleizer.py scripts/articles/editing_prompt.txt /home/mike/repos/trimnoir/_config.yml /home/mike/repos/grimoire/_config.yml ! rg -c 'futureproof' --glob '!AI_CONTEXT.md' . | sort ! for f in "$HOME/repos/nixos/blogs.nix" scripts/articles/lsa.py scripts/articles/generate_ai_context.py scripts/articles/common.py; do printf '%3s %s\n' "$(grep -c permalink_prefix "$f")" "$f"; done ! jq -r 'to_entries[] | "\(.key) permalink_prefix=\(.value.permalink_prefix // "MISSING")"' "$HOME/.config/pipulate/blogs.json" ! .venv/bin/python -c "import sys; sys.path.insert(0,'scripts/articles'); import lsa; [print(k, lsa.default_permalink('demo', lsa.permalink_prefix(v))) for k, v in sorted(lsa.load_targets().items())]" ! grep -c "get('url'" scripts/articles/generate_llms_txt.py ! grep -n permalink "$HOME/repos/trimnoir/_config.yml" "$HOME/repos/grimoire/_config.yml" 2>&1 | head -10 ! posts -t 1 --fmt dated-slugs --last 3 ``` ## 3. PATCHES **Car 1 — `blogs.nix` declares the folder.** Four blocks, one commit. Every SEARCH is a `preview_port` line, which is unique per target. **Target 4's `confluence_parent_id` line is deliberately not quoted anywhere below**: it arrives here as `[REDACTED_PARENT_ID]` from the compile-lane substitution table, so it is sanitized text, not raw source, and a SEARCH block containing it would fail the exact-match interlock with a diagnostic that looks like an indentation error and is not one. The `preview_port = 4004;` anchor sits above it and never touches it. ```text Target: /home/mike/repos/nixos/blogs.nix [[[SEARCH]]] preview_port = 4001; [[[DIVIDER]]] # THE BLOG FOLDER, DECLARED RATHER THAN HARDWIRED (2026-09-04). # Every consumer that has to INVENT a permalink -- lsa.py's # dated-slugs fallback, generate_ai_context.py's URL builder and # its two header strings -- carried the literal "futureproof", so # a second blog could not have a second folder no matter what it # declared. One value here, materialized into blogs.json and read # at call time, ends that. BARE SEGMENT, NO SLASHES: consumers add # those, and an empty string means posts live at the site root # rather than yielding a doubled slash. # ALL FOUR START AT "futureproof" ON PURPOSE. This car moves the # MECHANISM and changes no behavior; changing any one of these # values is a separate one-line commit with its own receipt. # NOT YET READ BY: editing_prompt.txt, which is what instructs the # model to write "/futureproof/[slug]/" into frontmatter in the # first place, and generate_post_nav.py, whose own fallback # disagrees with lsa.py's. Both are named; both are next. permalink_prefix = "futureproof"; preview_port = 4001; [[[REPLACE]]] ``` ```text Target: /home/mike/repos/nixos/blogs.nix [[[SEARCH]]] preview_port = 4002; [[[DIVIDER]]] permalink_prefix = "futureproof"; preview_port = 4002; [[[REPLACE]]] ``` ```text Target: /home/mike/repos/nixos/blogs.nix [[[SEARCH]]] preview_port = 4003; [[[DIVIDER]]] permalink_prefix = "futureproof"; preview_port = 4003; [[[REPLACE]]] ``` ```text Target: /home/mike/repos/nixos/blogs.nix [[[SEARCH]]] preview_port = 4004; [[[DIVIDER]]] permalink_prefix = "futureproof"; preview_port = 4004; [[[REPLACE]]] ``` **Car 2 — `lsa.py` grows the reader.** Three blocks, one commit. ```text Target: scripts/articles/lsa.py [[[SEARCH]]] def load_targets(): [[[DIVIDER]]] # THE BLOG FOLDER, READ AT CALL TIME (banked 2026-09-04). "futureproof" was a # literal in three files and a fourth disagreed with it, so a second blog could # not have a second folder no matter what it declared. The value now lives in # blogs.nix, materializes into blogs.json, and is read here. # LSA OWNS THE READER because it is the only module in this pipeline that BOTH # loads targets itself AND is imported by the standalone scripts, so a helper # here reaches every consumer without inventing a new import edge. # THE DEFAULT BELOW IS LOAD-BEARING, and it is also why a probe of these # helpers CANNOT witness a rebuild: a blogs.json written before the key existed # falls through to the default and prints exactly what a correctly-declared # target prints. Only a direct read of the JSON tells declared from defaulted. DEFAULT_PERMALINK_PREFIX = "futureproof" def permalink_prefix(target_config) -> str: """The blog's post folder, as a bare segment carrying no slashes. An explicitly declared empty string survives as empty -- that is a root-level blog, not a missing key -- which is why this tests for None rather than falsiness. """ raw = (target_config or {}).get('permalink_prefix') if raw is None: raw = DEFAULT_PERMALINK_PREFIX return str(raw).strip('/') def default_permalink(slug: str, prefix: str) -> str: """The permalink a post WOULD have if its frontmatter omitted one. No trailing slash, matching every call site, each of which appends "/index.md" itself. An empty prefix yields "/" rather than "//", so a root-level blog is representable, not a special case. """ return f"/{prefix}/{slug}" if prefix else f"/{slug}" def load_targets(): [[[REPLACE]]] ``` ```text Target: scripts/articles/lsa.py [[[SEARCH]]] base_url = target_config.get('base_url', 'https://mikelev.in').rstrip('/') [[[DIVIDER]]] base_url = target_config.get('base_url', 'https://mikelev.in').rstrip('/') prefix = permalink_prefix(target_config) [[[REPLACE]]] ``` ```text Target: scripts/articles/lsa.py [[[SEARCH]]] permalink = f"/futureproof/{slug}" [[[DIVIDER]]] permalink = default_permalink(slug, prefix) [[[REPLACE]]] ``` **Car 3 — `generate_ai_context.py` stops asserting a folder it does not know.** Twelve blocks, one commit. Two of them fix the header strings that tell an interrogating AI how to reconstruct every URL on the site; leaving those hardcoded while the code below them reads config is exactly the protection-lag shape this codebase keeps convicting. ```text Target: scripts/articles/generate_ai_context.py [[[SEARCH]]] def article_markdown_url(item: dict, base_url: str) -> str: [[[DIVIDER]]] def article_markdown_url(item: dict, base_url: str, prefix: str) -> str: [[[REPLACE]]] ``` ```text Target: scripts/articles/generate_ai_context.py [[[SEARCH]]] fall back to /futureproof//, always serving the raw index.md variant.""" [[[DIVIDER]]] fall back to the blog's declared permalink_prefix, always serving index.md.""" [[[REPLACE]]] ``` ```text Target: scripts/articles/generate_ai_context.py [[[SEARCH]]] permalink = f"/futureproof/{slug}" [[[DIVIDER]]] permalink = lsa.default_permalink(slug, prefix) [[[REPLACE]]] ``` ```text Target: scripts/articles/generate_ai_context.py [[[SEARCH]]] def build_header(article_count: int, base_url: str) -> str: [[[DIVIDER]]] def build_header(article_count: int, base_url: str, prefix: str) -> str: [[[REPLACE]]] ``` ```text Target: scripts/articles/generate_ai_context.py [[[SEARCH]]] host = base_url.replace("https://", "").replace("http://", "").rstrip("/") [[[DIVIDER]]] host = base_url.replace("https://", "").replace("http://", "").rstrip("/") folder = f"/{prefix}" if prefix else "" [[[REPLACE]]] ``` ```text Target: scripts/articles/generate_ai_context.py [[[SEARCH]]] URL as: `{base_url}/futureproof/{{slug}}/index.md` [[[DIVIDER]]] URL as: `{base_url}{folder}/{{slug}}/index.md` [[[REPLACE]]] ``` ```text Target: scripts/articles/generate_ai_context.py [[[SEARCH]]] https://{host}/futureproof/magic-cookie-pattern-self-bootstrapping-sovereignty/index.md [[[DIVIDER]]] https://{host}{folder}/magic-cookie-pattern-self-bootstrapping-sovereignty/index.md [[[REPLACE]]] ``` ```text Target: scripts/articles/generate_ai_context.py [[[SEARCH]]] metadata = lsa.get_holographic_article_data(str(target_path)) # newest-first [[[DIVIDER]]] metadata = lsa.get_holographic_article_data(str(target_path)) # newest-first prefix = lsa.permalink_prefix(target_config) folder = f"/{prefix}" if prefix else "" url_pattern = f"{base_url}{folder}/{{slug}}/index.md" [[[REPLACE]]] ``` ```text Target: scripts/articles/generate_ai_context.py [[[SEARCH]]] url = article_markdown_url(item, base_url) [[[DIVIDER]]] url = article_markdown_url(item, base_url, prefix) [[[REPLACE]]] ``` ```text Target: scripts/articles/generate_ai_context.py [[[SEARCH]]] f"\n## Compact slug index — pattern: {base_url}/futureproof/{{slug}}/index.md\n" f"\nFormat: `[date] [size] slug` — fetch any entry as `{base_url}/futureproof/{{slug}}/index.md`\n" [[[DIVIDER]]] f"\n## Compact slug index — pattern: {url_pattern}\n" f"\nFormat: `[date] [size] slug` — fetch any entry as `{url_pattern}`\n" [[[REPLACE]]] ``` ```text Target: scripts/articles/generate_ai_context.py [[[SEARCH]]] limit = args.limit if (args.limit and args.limit > 0) else None [[[DIVIDER]]] limit = args.limit if (args.limit and args.limit > 0) else None prefix = lsa.permalink_prefix(target_config) [[[REPLACE]]] ``` ```text Target: scripts/articles/generate_ai_context.py [[[SEARCH]]] header = build_header(count, base_url) [[[DIVIDER]]] header = build_header(count, base_url, prefix) [[[REPLACE]]] ``` **Car 4 — retire the second name.** One block. `permalink_style` has never been read by anything; two names for one concept in the fallback path is how a later reader gets the wrong one. ```text Target: scripts/articles/common.py [[[SEARCH]]] "permalink_style": "/futureproof/:slug/" [[[DIVIDER]]] "permalink_prefix": "futureproof" [[[REPLACE]]] ``` **Car 5 — the key-drift fix, adjacent and optional.** Skip it if you want this turn to be purely about the folder; the probe on line 5 covers it either way, and it is a behavioral no-op today because target 1 is the only target with `generate_llms_txt.py` in its pipeline and its `base_url` equals the hardcoded fallback. ```text Target: scripts/articles/generate_llms_txt.py [[[SEARCH]]] base_url = target_config.get('url', 'https://mikelev.in') # Default fallback [[[DIVIDER]]] # KEY DRIFT, convicted by reading (2026-09-04): blogs.json declares # base_url and has never declared url, so this .get always fell through to # the mikelev.in default and every target's llms.txt was built with target # 1's hostname. generate_ai_context.get_base_url already tolerates both # spellings, which is the tell that the older name existed once. # WITNESSED STRUCTURALLY ONLY (the grep goes 1 -> 0). No behavioral receipt # exists today because target 1 is the sole target running this script and # its base_url equals the fallback, so the fix is a no-op until a second # target adopts the pipeline -- which is exactly when it would have bitten. base_url = (target_config.get('base_url') or target_config.get('url') or 'https://mikelev.in').rstrip('/') [[[REPLACE]]] ``` **Ignition.** Car 1 requires it, cars 2 through 5 do not. - Cars 2–5 edit Python that every probe imports or executes at call time. No ignition; the compile lane self-ignites. - **Car 1 does nothing until `blogs.json` is re-materialized.** Ignition is `cd ~/repos/nixos && sudo nixos-rebuild switch`. Without it, probe line 3 reads `MISSING` ×4 — which is the whole point of having it. **Choreography, with one wrinkle.** `blogs.nix` lives in a different repo, so `d`, `m`, and `blast` from the pipulate root will not see it. The Target line is absolute, so `patch` then `app` from the pipulate root edits the right file; the commit is not. Run: ```text patch, app # Car 1 -- writes ~/repos/nixos/blogs.nix cd ~/repos/nixos && git add -A && git commit && cd - sudo nixos-rebuild switch # Car 1's ignition patch, app, d, m # Cars 2, 3, 4, 5 -- each its own commit blast ``` ## 4. PROMPT ```text Five cars moved the blog folder out of the code and into blogs.nix as permalink_prefix. Grade the LIVE RECEIPTS: census -- generate_ai_context.py drops off entirely (6 -> 0), lsa.py goes 1 -> 2 with its remaining hits being only the default constant and a migration comment, common.py stays 1, editing_prompt.txt stays 2; counts 0 0 0 0 -> 4 3 3 1 for blogs.nix, lsa.py, generate_ai_context.py, common.py; jq MISSING x4 -> futureproof x4, and this is the ONLY probe that witnesses the nixos-rebuild ignition, because the Python helper reads through DEFAULT_PERMALINK_PREFIX and prints the same four lines either way; helper probe AttributeError -> four "N /futureproof/demo" lines; get('url' 1 -> 0; posts --fmt dated-slugs byte-identical modulo the cold-cache stderr line, which proves no regression and NOT the new path, since every real post carries a frontmatter permalink and never reaches the fallback. If the census names a file that was not in the previous context, report it and patch nothing. Then land the next step, with _config.yml now in context for both targets. (a) Template the folder into editing_prompt.txt: articleizer.py already resolves target_config and already does one placeholder substitution ([INSERT BOOK SPINE]) before the API call, so add [INSERT PERMALINK PREFIX] the same way and stop instructing every blog to write "/futureproof/[slug]/" into its frontmatter. Two sites in that file need it: the permalink construction rule and the promotional-tweet URL example. (b) Rule on generate_post_nav.py's fallback, citing _config.yml's permalink setting rather than guessing: it returns "//" where lsa.default_permalink returns "//", and one of them is wrong about what Jekyll actually serves for a permalink-less post. If _config.yml turns out not to declare permalink at all, say so and name Jekyll's default rather than inventing one. Also standing, not for this turn: three more hardwired blog specifics are visible in the existing context and want their own car later -- build_knowledge_graph.py hardcodes base_url="https://mikelev.in" as a default parameter in project_llms_txt and project_xml_sitemaps and both are called without the argument, plus a root node titled "MikeLev.in" with a hardcoded blurb; generate_hubs.py hardcodes the same base_url in its JSON-LD breadcrumb builder; gsc_historical_fetch.py hardcodes SITE_URL = "sc-domain:mikelev.in". Each is the same disease as permalink_prefix and each needs a decision about which key it reads, not just where the string moves. ``` ## 5. EXTERNAL DELIVERABLES None this turn. Every change lands in `~/repos/nixos` or `~/repos/pipulate`, and the only out-of-band action is the `nixos-rebuild switch` named as Car 1's ignition. **MikeLev.in**: > Same commands, run twice, one change between them. Where the readings > differ is what the change did; the diff in the middle is the receipt. **1: Probe** (BEFORE: hand-run, nothing changed yet): ```bash $ git status On branch main Your branch is up to date with 'origin/main'. nothing to commit, working tree clean (nix) pipulate $ rg -c 'futureproof' --glob '!AI_CONTEXT.md' . | sort for f in "$HOME/repos/nixos/blogs.nix" scripts/articles/lsa.py scripts/articles/generate_ai_context.py scripts/articles/common.py; do printf '%3s %s\n' "$(grep -c permalink_prefix "$f")" "$f"; done jq -r 'to_entries[] | "\(.key) permalink_prefix=\(.value.permalink_prefix // "MISSING")"' "$HOME/.config/pipulate/blogs.json" .venv/bin/python -c "import sys; sys.path.insert(0,'scripts/articles'); import lsa; [print(k, lsa.default_permalink('demo', lsa.permalink_prefix(v))) for k, v in sorted(lsa.load_targets().items())]" grep -c "get('url'" scripts/articles/generate_llms_txt.py grep -n permalink "$HOME/repos/trimnoir/_config.yml" "$HOME/repos/grimoire/_config.yml" 2>&1 | head -10 posts -t 1 --fmt dated-slugs --last 3 ./assets/axis_ledger.jsonl:1 ./assets/prompts/find404s.md:1 ./foo_files.py:10 ./remotes/honeybot/queries/hot_404_remaps_top.sql:1 ./remotes/honeybot/queries/hub_candidates.sql:2 ./scripts/articles/common.py:1 ./scripts/articles/editing_prompt.txt:2 ./scripts/articles/generate_ai_context.py:6 ./scripts/articles/lsa.py:1 ./scripts/gsc/gsc_top_movers.py:1 0 /home/mike/repos/nixos/blogs.nix 0 scripts/articles/lsa.py 0 scripts/articles/generate_ai_context.py 0 scripts/articles/common.py 1 permalink_prefix=MISSING 2 permalink_prefix=MISSING 3 permalink_prefix=MISSING 4 permalink_prefix=MISSING Traceback (most recent call last): File "", line 1, in AttributeError: module 'lsa' has no attribute 'default_permalink' 1 /home/mike/repos/trimnoir/_config.yml:34:permalink: /:slug/ 2026-09-04 [ 43.3k Σ 43.3k] https://mikelev.in/futureproof/forgetting-with-style-replayable-memory-engine/index.md 2026-09-04 [ 23.9k Σ 67.2k] https://mikelev.in/futureproof/closing-the-loop-replayable-ai-workflows-and-manual-edits/index.md 2026-09-04 [ 53.8k Σ 121.0k] https://mikelev.in/futureproof/blind-fanout-model-evaluation-and-flight-recorder/index.md # ── selection: 3 articles | 121,039 tokens | 448,993 bytes (Σ121.0k) (nix) pipulate $ ``` **2: Context** (AFTER: the same probes re-run by the compiler as `!` lines): ```text # adhoc.txt _ _ _ to set context____ _ _ ___ ____ _ Simpson Couch Gag Here (explain anything to the audience you feel needs it explained)G # / \ __| | | | | | ___ ___ / ___| | | |/ _ \| _ \| | # ahe/ _ \ / _` | | |_| |/ _ \ / __| | | | |_| | | | | |_) | | Making the blogging system less hardwired to a particular subfolder. # ahc ___ \ (_| | | _ | (_) | (__ | |___| _ | |_| | __/|_| # /_/ \_\__,_| |_| |_|\___/ \___| \____|_| |_|\___/|_| (_) # Ad Hoc CHOP: The Not-Managed-by-Git Safe-for-Client-Data place # OPTIONAL BUT BIG FOR FULL CONTEXT-WINDOW STORYTELLING # ! python scripts/articles/lsa.py -t 1 --reverse --fmt dated-slugs # <-- The "Rolling Pin" that gives the 40K foot book-spine view of book-ore. # GLOSSARY.md # <-- Like the back of a J.R.R. Tolkien book but always growing in size as `prompt_foo.py` gets scars and shrinks. # scripts/articles/lsa.py # <-- 2nd Brain query-engine for `rgx`, `rgxc` & `posts` Jekyll-inspired Memory Externalization for Hackers. # ~/repos/nixos/autognome.py # <-- Letting the AIs really understand my environment (The Brave Little Tailor punches above Their Weight Class proving the dunning-kruger effect the gate-keeper's (lower-case) lament.) init.lua # <-- Daily driver hot-keys that overlap with aliases in flake.nix. `m` makes it Science (this process)! # STILL BIG BUT LESS OPTIONAL (especially flake.nix) flake.nix # <-- THE ONE BIG THING TO INCLUDE Infrastructure as Code (IaC) tells LLM about your system down to the metal # prompt_foo.py # <-- This very content-compiling system # foo_files.py # <-- This is the router, evolving book outline and the things you pin-up to produced the recursive self-improvement loops # TINY ILLUMINATING (OK to include every time / automatically = `apply.py`, `.gitignore`, `.gitattributes`) # requirements.in # <-- All known dependencies and (necessary) version pinning. WORA gotcha's exposed. # __init__.py # <-- Master versioning # pyproject.toml # <-- The PyPI Packaging details # OPTIONAL ACTUATORS (cheap and good to include to expand the AI's capabilities) # cli.py # <-- Catch-all actuator for PyPI envs, Python anchoring, MCP tool-call (plus alternatives) and **kwargs like wrapping for CLI # scripts/xp.py # <-- Transforms host OS copy-paste buffer player-piano music into context-payload. # scripts/ai.py # <-- How I constantly use local AI to write git commit messages with `m` alias. # scripts/crawl.py # <-- Feel free to ask for something to be crawled and included in the next turn. # scripts/weblogin.py # <-- Lets the user "warm up" the cache for their web logins at their leisure on a profile that persists. # MISCELLANEOUS (rare to include but sometimes critical) # scripts/foo_cartridge.py # Needs description # scripts/foo_replay.py # Needs description # release.py # <-- How everything ends up where it does (GitHub, PyPI, etc.) # imports/voice_synthesis.py # <-- The wand can talk to you # imports/ascii_displays.py # <-- Where all the ASCII Art lives # scripts/release/version_sync.py # <-- Needs to be wrapped into release.py and eliminated, I think. # --- Under this line is were you paste what the AI gives you --- # --- We call it context but it's really just the right-hand --- # --- blast-radius of the "probes" to make this all science. --- # Carry-over as the important work-in-progress parts of the project here just # like above but not as long-standing overarching to the framework but rather # for the current hot spots actively being worked on. # STICKBUG & MOTHER CAT KATA (WORKING ON THE CHAPTER) # assets/installer/mck.sh # assets/installer/replay.sh # scripts/walk.py # scripts/walk_cartridge.py # scripts/walk_compile.py # assets/trails/first_context.yaml # assets/trails/practice.yaml # assets/trails/public_walk.yaml # assets/trails/botify_pageworkers.yaml # scripts/connectors/README.md # scripts/connectors/botify.py # scripts/connectors/confluence.py # scripts/connectors/gmail.py # scripts/connectors/gsc.py # scripts/connectors/jira.py # scripts/connectors/mcp.py # scripts/connectors/mcp_warm.py # scripts/connectors/sheets.py # scripts/connectors/slack.py # scripts/connectors/wallet.py # scripts/bookmark_import.py # scripts/boot_menu.py # scripts/mother_cat.py # scripts/sources_menu.py # scripts/weblogin.py # tools/scraper_tools.py # --- START THIS DISCUSSION --- # Context 1 (Edit-in selections from above and add new files immediately below) # # /home/mike/repos/grimoire/_posts/2026-09-03-three-door-boot-menu-reproducible-startup.md # /home/mike/repos/trimnoir/_posts/2026-09-04-closing-the-loop-replayable-ai-workflows-and-manual-edits.md # # # CORE SLIDESHOW FILES # nixops.sh # <-- You've heard of GitOPs? Well, this is NixOPs. # remotes/honeybot/hooks/post-receive # <-- Ever hear of GitHub Pages? Or github.io? This is that. # remotes/honeybot/nixos/configuration.nix # <-- It's as if Pipulate had kids. Spy kids. # remotes/honeybot/scripts/stream.py # <-- Starts the TV Channel streaming to YouTube-live via OBS from Nginx Honeybot XFCE Desktop. Clear? # remotes/honeybot/scripts/score.py # <-- Where "Greetings Entity" slideshow reads on post-receive interrupts # remotes/honeybot/scripts/card.py # <-- Just added for station identification breaks # remotes/honeybot/scripts/forest.py # <-- Likewise, just added for the new storytelling system on Honeybot # remotes/honeybot/scripts/test_forest.py # <-- Test Honeybot station identification sequence on Pipulate Prime # remotes/honeybot/scripts/logs.py # <-- The TV Show is mostly Nginx `access.log` files tailed and piped through Python to colorize (this). # remotes/honeybot/scripts/content_loader.py # <-- Tricky TV programming & scheduling stuff. Absolute versus relative timing. Loops. Interrupts. # remotes/honeybot/scripts/db.py # <-- But you can't keep your weblogs forever! And we want trending. And data-mining. Here's how. # imports/voice_synthesis.py # <-- The wand can talk to you (not sure if I'm keeping it in Honeybot chapter) # # # JEKYLL PUBLISHING # ~/.config/pipulate/blogs.json # <-- CAUTION! Derived from ~/repos/nixos/blogs.nix # scripts/articles/publishizer.py # <-- Orchestrates different publishing workflows per target blog. # scripts/articles/common.py # <-- Self-explanatory # scripts/articles/articleizer.py # <-- Transforms raw article.txt to formal Jekyll markdown format # scripts/articles/editing_prompt.txt # <-- Forcing response into strict JSON data structure # scripts/articles/sanitizer.py # <-- Scrubs PII # scripts/articles/gsc_historical_fetch.py # scripts/articles/contextualizer.py # <-- Builds JSON summaries of articles in `_posts/context/` called "Holographic Shards". # scripts/articles/confluenceizer.py # <-- Idempotent Jekyll-to-Confluence corporate wiki # scripts/articles/googledocizer.py # <-- Just added # scripts/articles/build_knowledge_graph.py # <-- Topically load-balances site using hierarchical K-Means keyword clustering groups # scripts/articles/generate_ai_context.py # <-- AIs WILL interrogate your repo. This gives epic context of article URLs for drill-down. # scripts/articles/generate_hubs.py # <-- Uses just-produced link-graph data to generate each of the new hubs it suggests # scripts/articles/generate_llms_txt.py # <-- Builds an llms.txt based on the auto-organized structure suggested here # scripts/articles/generate_redirects.py # <-- Generates redirect map above hub-churn suggests is needed # scripts/articles/sanitize_redirects.py # <-- Deals with follow-up meticulous pedantic detail required for a good Nginx redirect map # # /home/mike/repos/trimnoir/_layouts/default.html # /home/mike/repos/trimnoir/_layouts/forward.html # /home/mike/repos/trimnoir/_layouts/home.html # /home/mike/repos/trimnoir/_layouts/page.html # /home/mike/repos/trimnoir/_layouts/plain.html # /home/mike/repos/trimnoir/_layouts/post.html # /home/mike/repos/grimoire/flake.nix # /home/mike/repos/grimoire/_layouts/default.html # /home/mike/repos/grimoire/index.md # # /home/mike/repos/nixos/blogs.nix # /home/mike/.config/pipulate/blogs.json # scripts/articles/lsa.py # scripts/articles/link_injector.py # scripts/articles/generate_post_nav.py # Context 2 /home/mike/repos/nixos/blogs.nix /home/mike/.config/pipulate/blogs.json scripts/articles/lsa.py scripts/articles/generate_ai_context.py scripts/articles/common.py scripts/articles/generate_llms_txt.py scripts/articles/generate_post_nav.py scripts/articles/articleizer.py scripts/articles/editing_prompt.txt /home/mike/repos/trimnoir/_config.yml /home/mike/repos/grimoire/_config.yml ! rg -c 'futureproof' --glob '!AI_CONTEXT.md' . | sort ! for f in "$HOME/repos/nixos/blogs.nix" scripts/articles/lsa.py scripts/articles/generate_ai_context.py scripts/articles/common.py; do printf '%3s %s\n' "$(grep -c permalink_prefix "$f")" "$f"; done ! jq -r 'to_entries[] | "\(.key) permalink_prefix=\(.value.permalink_prefix // "MISSING")"' "$HOME/.config/pipulate/blogs.json" ! .venv/bin/python -c "import sys; sys.path.insert(0,'scripts/articles'); import lsa; [print(k, lsa.default_permalink('demo', lsa.permalink_prefix(v))) for k, v in sorted(lsa.load_targets().items())]" ! grep -c "get('url'" scripts/articles/generate_llms_txt.py ! grep -n permalink "$HOME/repos/trimnoir/_config.yml" "$HOME/repos/grimoire/_config.yml" 2>&1 | head -10 ! posts -t 1 --fmt dated-slugs --last 3 # --- END `adhoc.txt` TEMPLATE --- ``` **3: Patches** (the one change between the readings): ```bash Blast Radius Check to establish bisection Left-hand Causal Boundary. It is a Popper-thing. Science. On branch main Your branch is up to date with 'origin/main'. nothing to commit, working tree clean (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated '/home/mike/repos/nixos/blogs.nix'. (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated '/home/mike/repos/nixos/blogs.nix'. (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated '/home/mike/repos/nixos/blogs.nix'. (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated '/home/mike/repos/nixos/blogs.nix'. (nix) pipulate $ ``` And that did this: ```diff (sys) nixos $ git --no-pager diff -- blogs.nix diff --git a/blogs.nix b/blogs.nix index 290354b..faa955e 100644 --- a/blogs.nix +++ b/blogs.nix @@ -22,6 +22,23 @@ let lane = "public"; alias = "article"; base_url = "https://mikelev.in"; + # THE BLOG FOLDER, DECLARED RATHER THAN HARDWIRED (2026-09-04). + # Every consumer that has to INVENT a permalink -- lsa.py's + # dated-slugs fallback, generate_ai_context.py's URL builder and + # its two header strings -- carried the literal "futureproof", so + # a second blog could not have a second folder no matter what it + # declared. One value here, materialized into blogs.json and read + # at call time, ends that. BARE SEGMENT, NO SLASHES: consumers add + # those, and an empty string means posts live at the site root + # rather than yielding a doubled slash. + # ALL FOUR START AT "futureproof" ON PURPOSE. This car moves the + # MECHANISM and changes no behavior; changing any one of these + # values is a separate one-line commit with its own receipt. + # NOT YET READ BY: editing_prompt.txt, which is what instructs the + # model to write "/futureproof/[slug]/" into frontmatter in the + # first place, and generate_post_nav.py, whose own fallback + # disagrees with lsa.py's. Both are named; both are next. + permalink_prefix = "futureproof"; preview_port = 4001; # Drive folder googledocizer.py upserts into. Bootstrap once with: # python scripts/articles/googledocizer.py --bootstrap-folder "MikeLev.in Articles" @@ -45,6 +62,7 @@ let path = "/home/mike/repos/Pipulate.com/_posts"; lane = "public"; base_url = "https://pipulate.com"; + permalink_prefix = "futureproof"; preview_port = 4002; pipeline = [ "contextualizer.py" ]; }; @@ -54,6 +72,7 @@ let lane = "private"; alias = "grim"; base_url = "http://nixos.local:4003"; + permalink_prefix = "futureproof"; preview_port = 4003; pipeline = [ "contextualizer.py" ]; }; @@ -63,6 +82,7 @@ let lane = "public"; alias = "bot"; base_url = ""; + permalink_prefix = "futureproof"; preview_port = 4004; confluence_parent_id = "[REDACTED_PARENT_ID]"; pipeline = [ "contextualizer.py" "confluenceizer.py" ]; (sys) nixos $ git commit -am "Making blogs.nix support subfolders" [main 5c874f2] Making blogs.nix support subfolders 3 files changed, 23 insertions(+) (sys) nixos $ git push Enumerating objects: 9, done. Counting objects: 100% (9/9), done. Delta compression using up to 48 threads Compressing objects: 100% (5/5), done. Writing objects: 100% (5/5), 2.69 KiB | 2.69 MiB/s, done. Total 5 (delta 4), reused 0 (delta 0), pack-reused 0 (from 0) remote: Resolving deltas: 100% (4/4), completed with 4 local objects. To github.com:miklevin/nixos-config.git 22041a5..5c874f2 main -> main (sys) nixos $ ``` And the next car: ```diff Blast Radius Check to establish bisection Left-hand Causal Boundary. It is a Popper-thing. Science. On branch main Your branch is up to date with 'origin/main'. nothing to commit, working tree clean (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/lsa.py'. (nix) pipulate $ d diff --git a/scripts/articles/lsa.py b/scripts/articles/lsa.py index cf372a80..3ac85ec3 100644 --- a/scripts/articles/lsa.py +++ b/scripts/articles/lsa.py @@ -109,6 +109,43 @@ class MtimeMemo: except Exception: pass +# THE BLOG FOLDER, READ AT CALL TIME (banked 2026-09-04). "futureproof" was a +# literal in three files and a fourth disagreed with it, so a second blog could +# not have a second folder no matter what it declared. The value now lives in +# blogs.nix, materializes into blogs.json, and is read here. +# LSA OWNS THE READER because it is the only module in this pipeline that BOTH +# loads targets itself AND is imported by the standalone scripts, so a helper +# here reaches every consumer without inventing a new import edge. +# THE DEFAULT BELOW IS LOAD-BEARING, and it is also why a probe of these +# helpers CANNOT witness a rebuild: a blogs.json written before the key existed +# falls through to the default and prints exactly what a correctly-declared +# target prints. Only a direct read of the JSON tells declared from defaulted. +DEFAULT_PERMALINK_PREFIX = "futureproof" + + +def permalink_prefix(target_config) -> str: + """The blog's post folder, as a bare segment carrying no slashes. + + An explicitly declared empty string survives as empty -- that is a + root-level blog, not a missing key -- which is why this tests for None + rather than falsiness. + """ + raw = (target_config or {}).get('permalink_prefix') + if raw is None: + raw = DEFAULT_PERMALINK_PREFIX + return str(raw).strip('/') + + +def default_permalink(slug: str, prefix: str) -> str: + """The permalink a post WOULD have if its frontmatter omitted one. + + No trailing slash, matching every call site, each of which appends + "/index.md" itself. An empty prefix yields "/" rather than + "//", so a root-level blog is representable, not a special case. + """ + return f"/{prefix}/{slug}" if prefix else f"/{slug}" + + def load_targets(): if TARGETS_FILE.exists(): try: (nix) pipulate $ m 📝 Committing: chore: Update permalink prefix handling in lsa.py [main 3ac4940b] chore: Update permalink prefix handling in lsa.py 1 file changed, 37 insertions(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/lsa.py'. (nix) pipulate $ d diff --git a/scripts/articles/lsa.py b/scripts/articles/lsa.py index 3ac85ec3..0114079b 100644 --- a/scripts/articles/lsa.py +++ b/scripts/articles/lsa.py @@ -571,6 +571,7 @@ def main(): import re target_config = targets[target_key] base_url = target_config.get('base_url', 'https://mikelev.in').rstrip('/') + prefix = permalink_prefix(target_config) # THE BUDGET MAP: per-article token size PLUS a running Σ cumulative. # Newest-first (--reverse) means Σ answers "take the N newest -> this # many tokens." Scan the Σ column top-down; cut where it crosses your (nix) pipulate $ m 📝 Committing: chore: Update permalink prefix logic in lsa.py [main e5edb0c4] chore: Update permalink prefix logic in lsa.py 1 file changed, 1 insertion(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/lsa.py'. (nix) pipulate $ d diff --git a/scripts/articles/lsa.py b/scripts/articles/lsa.py index 0114079b..0d74e2b9 100644 --- a/scripts/articles/lsa.py +++ b/scripts/articles/lsa.py @@ -587,7 +587,7 @@ def main(): # Leverages YAML frontmatter permalinks falling back to default route structures. permalink = item.get('permalink', '').rstrip('/') if not permalink: - permalink = f"/futureproof/{slug}" + permalink = default_permalink(slug, prefix) url_target = f"{base_url}{permalink}/index.md" if tokens > 0: run_tokens += tokens (nix) pipulate $ m 📝 Committing: chore: Update default permalink generation in lsa.py [main 76fc1410] chore: Update default permalink generation in lsa.py 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ git push Enumerating objects: 19, done. Counting objects: 100% (19/19), done. Delta compression using up to 48 threads Compressing objects: 100% (15/15), done. Writing objects: 100% (15/15), 2.23 KiB | 2.23 MiB/s, done. Total 15 (delta 12), reused 0 (delta 0), pack-reused 0 (from 0) remote: Resolving deltas: 100% (12/12), completed with 4 local objects. To github.com:pipulate/pipulate.git a28aebf8..76fc1410 main -> main (nix) pipulate $ ``` I'm thinking about breaking up documenting these patches based on the so-called Cars. ## Applying Patch Cars Across Codebase Boundaries ```diff (nix) pipulate $ g Blast Radius Check to establish bisection Left-hand Causal Boundary. It is a Popper-thing. Science. On branch main Your branch is up to date with 'origin/main'. nothing to commit, working tree clean (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_ai_context.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_ai_context.py b/scripts/articles/generate_ai_context.py index 7a1845bd..0e4df195 100644 --- a/scripts/articles/generate_ai_context.py +++ b/scripts/articles/generate_ai_context.py @@ -49,7 +49,7 @@ def get_base_url(target_config: dict) -> str: return (target_config.get("base_url") or target_config.get("url") or DEFAULT_BASE_URL).rstrip("/") -def article_markdown_url(item: dict, base_url: str) -> str: +def article_markdown_url(item: dict, base_url: str, prefix: str) -> str: """Mirror lsa.py --fmt dated-slugs routing: honor the YAML permalink, else fall back to /futureproof//, always serving the raw index.md variant.""" permalink = (item.get("permalink") or "").rstrip("/") (nix) pipulate $ m 📝 Committing: chore: Add prefix to article_markdown_url function signature [main d628c49c] chore: Add prefix to article_markdown_url function signature 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_ai_context.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_ai_context.py b/scripts/articles/generate_ai_context.py index 0e4df195..a4ad1127 100644 --- a/scripts/articles/generate_ai_context.py +++ b/scripts/articles/generate_ai_context.py @@ -51,7 +51,7 @@ def get_base_url(target_config: dict) -> str: def article_markdown_url(item: dict, base_url: str, prefix: str) -> str: """Mirror lsa.py --fmt dated-slugs routing: honor the YAML permalink, else - fall back to /futureproof//, always serving the raw index.md variant.""" + fall back to the blog's declared permalink_prefix, always serving index.md.""" permalink = (item.get("permalink") or "").rstrip("/") if not permalink: stem = Path(item["filename"]).stem (nix) pipulate $ m 📝 Committing: chore: Refactor article URL generation logic [main cbf012b1] chore: Refactor article URL generation logic 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_ai_context.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_ai_context.py b/scripts/articles/generate_ai_context.py index a4ad1127..cd0a1971 100644 --- a/scripts/articles/generate_ai_context.py +++ b/scripts/articles/generate_ai_context.py @@ -56,7 +56,7 @@ def article_markdown_url(item: dict, base_url: str, prefix: str) -> str: if not permalink: stem = Path(item["filename"]).stem slug = re.sub(r"^\d{4}-\d{2}-\d{2}-", "", stem) - permalink = f"/futureproof/{slug}" + permalink = lsa.default_permalink(slug, prefix) return f"{base_url}{permalink}/index.md" (nix) pipulate $ m 📝 Committing: chore: Update permalink generation in generate_ai_context.py [main 0ecce553] chore: Update permalink generation in generate_ai_context.py 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_ai_context.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_ai_context.py b/scripts/articles/generate_ai_context.py index cd0a1971..7edd44a4 100644 --- a/scripts/articles/generate_ai_context.py +++ b/scripts/articles/generate_ai_context.py @@ -69,7 +69,7 @@ def article_slug(item: dict) -> str: return re.sub(r"^\d{4}-\d{2}-\d{2}-", "", stem) -def build_header(article_count: int, base_url: str) -> str: +def build_header(article_count: int, base_url: str, prefix: str) -> str: """The static framing an interrogating AI reads before the index.""" today = datetime.now().strftime("%Y-%m-%d") host = base_url.replace("https://", "").replace("http://", "").rstrip("/") (nix) pipulate $ m 📝 Committing: chore: Add `prefix` parameter to `build_header` function [main 3766d388] chore: Add `prefix` parameter to `build_header` function 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_ai_context.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_ai_context.py b/scripts/articles/generate_ai_context.py index 7edd44a4..e046b241 100644 --- a/scripts/articles/generate_ai_context.py +++ b/scripts/articles/generate_ai_context.py @@ -73,6 +73,7 @@ def build_header(article_count: int, base_url: str, prefix: str) -> str: """The static framing an interrogating AI reads before the index.""" today = datetime.now().strftime("%Y-%m-%d") host = base_url.replace("https://", "").replace("http://", "").rstrip("/") + folder = f"/{prefix}" if prefix else "" return f"""# AI_CONTEXT.md — Start Here If You Are an AI > Auto-generated on {today}. Rewritten from scratch on every release. If this (nix) pipulate $ m 📝 Committing: chore: Refactor AI context header generation [main 2bde2a48] chore: Refactor AI context header generation 1 file changed, 1 insertion(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_ai_context.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_ai_context.py b/scripts/articles/generate_ai_context.py index e046b241..9949d2d1 100644 --- a/scripts/articles/generate_ai_context.py +++ b/scripts/articles/generate_ai_context.py @@ -170,7 +170,7 @@ For a guided, interactive tour of the codebase itself, run The first {FULL_URL_THRESHOLD} entries include full `index.md` URLs to establish the link pattern. All remaining entries are bare slugs. Reconstruct any full -URL as: `{base_url}/futureproof/{{slug}}/index.md` +URL as: `{base_url}{folder}/{{slug}}/index.md` """ (nix) pipulate $ m 0📝 Committing: fix: Update URL construction in generate_ai_context.py [main f437506c] fix: Update URL construction in generate_ai_context.py 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_ai_context.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_ai_context.py b/scripts/articles/generate_ai_context.py index 9949d2d1..8bd2ab58 100644 --- a/scripts/articles/generate_ai_context.py +++ b/scripts/articles/generate_ai_context.py @@ -155,7 +155,7 @@ recompiles a fresh context payload. [triple-backtick]text [[[TODO_SLUGS]]] deterministic-ai-wet-philosophy -https://{host}/futureproof/magic-cookie-pattern-self-bootstrapping-sovereignty/index.md +https://{host}{folder}/magic-cookie-pattern-self-bootstrapping-sovereignty/index.md [[[END_SLUGS]]] [[[TODO_FILES]]] (nix) pipulate $ m 📝 Committing: chore: Update AI context URL in generate_ai_context.py [main 8640d8f2] chore: Update AI context URL in generate_ai_context.py 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_ai_context.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_ai_context.py b/scripts/articles/generate_ai_context.py index 8bd2ab58..a5a64533 100644 --- a/scripts/articles/generate_ai_context.py +++ b/scripts/articles/generate_ai_context.py @@ -184,6 +184,9 @@ def build_ledger(target_config: dict, rich: bool, limit) -> tuple: return "", 0 metadata = lsa.get_holographic_article_data(str(target_path)) # newest-first + prefix = lsa.permalink_prefix(target_config) + folder = f"/{prefix}" if prefix else "" + url_pattern = f"{base_url}{folder}/{{slug}}/index.md" if limit: metadata = metadata[:limit] (nix) pipulate $ m 📝 Committing: chore: Update URL pattern in generate_ai_context.py [main 26ce19ab] chore: Update URL pattern in generate_ai_context.py 1 file changed, 3 insertions(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_ai_context.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_ai_context.py b/scripts/articles/generate_ai_context.py index a5a64533..f10f8e7f 100644 --- a/scripts/articles/generate_ai_context.py +++ b/scripts/articles/generate_ai_context.py @@ -200,7 +200,7 @@ def build_ledger(target_config: dict, rich: bool, limit) -> tuple: byte_size = 0 size_k = f"{max(1, round(byte_size / 1000))}k" if byte_size else "?" if idx < FULL_URL_THRESHOLD: - url = article_markdown_url(item, base_url) + url = article_markdown_url(item, base_url, prefix) line = f"- [{item['date']}] [{title}]({url})" else: if idx == FULL_URL_THRESHOLD: (nix) pipulate $ m 📝 Committing: chore: Update article\_markdown\_url with prefix [main 89556569] chore: Update article\_markdown\_url with prefix 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_ai_context.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_ai_context.py b/scripts/articles/generate_ai_context.py index f10f8e7f..70301da6 100644 --- a/scripts/articles/generate_ai_context.py +++ b/scripts/articles/generate_ai_context.py @@ -205,8 +205,8 @@ def build_ledger(target_config: dict, rich: bool, limit) -> tuple: else: if idx == FULL_URL_THRESHOLD: lines.append( - f"\n## Compact slug index — pattern: {base_url}/futureproof/{{slug}}/index.md\n" - f"\nFormat: `[date] [size] slug` — fetch any entry as `{base_url}/futureproof/{{slug}}/index.md`\n" + f"\n## Compact slug index — pattern: {url_pattern}\n" + f"\nFormat: `[date] [size] slug` — fetch any entry as `{url_pattern}`\n" ) slug = article_slug(item) line = f"- [{item['date']}] [{size_k}] {slug}" (nix) pipulate $ m 📝 Committing: chore: Update slug index pattern in generate_ai_context.py [main 9a751fb6] chore: Update slug index pattern in generate_ai_context.py 1 file changed, 2 insertions(+), 2 deletions(-) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_ai_context.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_ai_context.py b/scripts/articles/generate_ai_context.py index 70301da6..6cfa3359 100644 --- a/scripts/articles/generate_ai_context.py +++ b/scripts/articles/generate_ai_context.py @@ -232,6 +232,7 @@ def main(): target_config = targets[target_key] base_url = get_base_url(target_config) limit = args.limit if (args.limit and args.limit > 0) else None + prefix = lsa.permalink_prefix(target_config) print(f"🧭 Generating AI_CONTEXT.md from target: {target_config.get('name', target_key)}") ledger, count = build_ledger(target_config, args.rich, limit) (nix) pipulate $ m 📝 Committing: chore: Add permalink prefix to generate_ai_context.py [main a048b07f] chore: Add permalink prefix to generate_ai_context.py 1 file changed, 1 insertion(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_ai_context.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_ai_context.py b/scripts/articles/generate_ai_context.py index 6cfa3359..c1cb328f 100644 --- a/scripts/articles/generate_ai_context.py +++ b/scripts/articles/generate_ai_context.py @@ -236,7 +236,7 @@ def main(): print(f"🧭 Generating AI_CONTEXT.md from target: {target_config.get('name', target_key)}") ledger, count = build_ledger(target_config, args.rich, limit) - header = build_header(count, base_url) + header = build_header(count, base_url, prefix) body = ledger if ledger else "_No articles indexed (article source unavailable at generation time)._" final = header + "\n" + body + "\n" (nix) pipulate $ m 📝 Committing: chore: Update build_header with prefix [main c78dbc26] chore: Update build_header with prefix 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ git push Enumerating objects: 64, done. Counting objects: 100% (64/64), done. Delta compression using up to 48 threads Compressing objects: 100% (60/60), done. Writing objects: 100% (60/60), 5.95 KiB | 2.97 MiB/s, done. Total 60 (delta 47), reused 0 (delta 0), pack-reused 0 (from 0) remote: Resolving deltas: 100% (47/47), completed with 3 local objects. To github.com:pipulate/pipulate.git 76fc1410..c78dbc26 main -> main (nix) pipulate $ ``` Wow this is pretty epic. And we're only just up to Car 4: ```diff Blast Radius Check to establish bisection Left-hand Causal Boundary. It is a Popper-thing. Science. On branch main Your branch is up to date with 'origin/main'. nothing to commit, working tree clean (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/common.py'. (nix) pipulate $ d diff --git a/scripts/articles/common.py b/scripts/articles/common.py index 993190f7..1d907272 100644 --- a/scripts/articles/common.py +++ b/scripts/articles/common.py @@ -18,7 +18,7 @@ DEFAULT_TARGETS = { "name": "Trimnoir (Personal Journal)", "path": str(Path.home() / "repos" / "trimnoir" / "_posts"), "base_url": "https://mikelev.in", - "permalink_style": "/futureproof/:slug/" + "permalink_prefix": "futureproof" } } (nix) pipulate $ m 📝 Committing: chore: Update permalink prefix in common.py [main 1f69c16e] chore: Update permalink prefix in common.py 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ git push Enumerating objects: 9, done. Counting objects: 100% (9/9), done. Delta compression using up to 48 threads Compressing objects: 100% (5/5), done. Writing objects: 100% (5/5), 463 bytes | 463.00 KiB/s, done. Total 5 (delta 4), reused 0 (delta 0), pack-reused 0 (from 0) remote: Resolving deltas: 100% (4/4), completed with 4 local objects. To github.com:pipulate/pipulate.git c78dbc26..1f69c16e main -> main (nix) pipulate $ ``` And now optional Car 5; I'm not big on keeping the articles topically pure that's more of a aspirational thing. Pragmatism will win every time. ```diff Blast Radius Check to establish bisection Left-hand Causal Boundary. It is a Popper-thing. Science. On branch main Your branch is up to date with 'origin/main'. nothing to commit, working tree clean (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_llms_txt.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_llms_txt.py b/scripts/articles/generate_llms_txt.py index 84bff1a4..fa48368b 100644 --- a/scripts/articles/generate_llms_txt.py +++ b/scripts/articles/generate_llms_txt.py @@ -20,7 +20,17 @@ OUTPUT_FILE = REPO_ROOT / "scripts" / "articles" / "llms.txt" def build_payload(target_config: dict) -> str: """Generates the dense, reverse-chronological ledger using lsa.py logic.""" target_path = Path(target_config['path']).expanduser().resolve() - base_url = target_config.get('url', 'https://mikelev.in') # Default fallback + # KEY DRIFT, convicted by reading (2026-09-04): blogs.json declares + # base_url and has never declared url, so this .get always fell through to + # the mikelev.in default and every target's llms.txt was built with target + # 1's hostname. generate_ai_context.get_base_url already tolerates both + # spellings, which is the tell that the older name existed once. + # WITNESSED STRUCTURALLY ONLY (the grep goes 1 -> 0). No behavioral receipt + # exists today because target 1 is the sole target running this script and + # its base_url equals the fallback, so the fix is a no-op until a second + # target adopts the pipeline -- which is exactly when it would have bitten. + base_url = (target_config.get('base_url') or target_config.get('url') + or 'https://mikelev.in').rstrip('/') print(f"📚 Extracting metadata from: {target_path}") # Leverage the Universal Semantic Extractor from lsa.py (nix) pipulate $ m 📝 Committing: chore: Refactor base_url handling in generate_llms_txt.py [main d189f097] chore: Refactor base_url handling in generate_llms_txt.py 1 file changed, 11 insertions(+), 1 deletion(-) (nix) pipulate $ git push Enumerating objects: 9, done. Counting objects: 100% (9/9), done. Delta compression using up to 48 threads Compressing objects: 100% (5/5), done. Writing objects: 100% (5/5), 952 bytes | 952.00 KiB/s, done. Total 5 (delta 4), reused 0 (delta 0), pack-reused 0 (from 0) remote: Resolving deltas: 100% (4/4), completed with 4 local objects. To github.com:pipulate/pipulate.git 1f69c16e..d189f097 main -> main (nix) pipulate $ ``` Ignition: Okay, Car 1 requires ignition because anything that touches a dot nix file needs the system to be rebuild and for those scratching their heads wondering what I'm keeping Jekyll blog configuration in Nix system files is about, it's about the great integrated life where you keep 1 great big single text-file for life so you never have to think about where to type just like this kind of prompt and articles and even just thinking out loud thing ever again in your life — you have a formal place for it now in `journal.txt`, but you've got all these other blogs that you can route your different ideas into as you go for publishing with different tracks, which in my case are: 1. A public personal blog (this) 2. A public blog related to a project (Pipulate.com / empty right now) 3. A private personal blog 4. A private professional for-work (full of proprietary stuff) blog So the upshot of all this is a sort of fast and furious writing about any and all things in a linear reverse-chronological way where very new thing at the top pushes everything out, and then you route your ideas where they go to "activate" them in the world (or not). This is like riding a bicycle and after sever months of this you will be able to publish (or not) at what seems like an inhuman rate, and all really in a way that pushes towards real meaningful *progress* — doing the right things in the right way statistically on average over time even though there will be plenty of mistakes along the way. But even those mistakes you capture and they become "reverse assets" showing the scars and the way things are the way they are which is just as valuable as what makes the cut of the fully baked deliverables. ## System Rebuilds and the Single Textfile Philosophy ```bash Blast Radius Check to establish bisection Left-hand Causal Boundary. It is a Popper-thing. Science. On branch main Your branch is up to date with 'origin/main'. nothing to commit, working tree clean (nix) pipulate $ n building Nix... building the system configuration... unpacking 'https://github.com/NixOS/nixpkgs/archive/nixos-unstable.tar.gz' into the Git cache... this derivation will be built: /nix/store/lxnx572acx74ay64nykav1zdyw7dnxvq-nixos-system-nixos-25.05.813814.ac62194c3917.drv building '/nix/store/lxnx572acx74ay64nykav1zdyw7dnxvq-nixos-system-nixos-25.05.813814.ac62194c3917.drv'... activating the configuration... setting up /etc... reloading user units for mike... restarting sysinit-reactivation.target the following new units were started: NetworkManager-dispatcher.service Done. The new configuration is /nix/store/kx9ifs7ly92l54zmc82bqjlhkckkajb1-nixos-system-nixos-25.05.813814.ac62194c3917 (nix) nixos $ p (nix) pipulate $ ``` There! **4: Prompt** (the next question, written before the AFTER reading is in): Five cars moved the blog folder out of the code and into blogs.nix as permalink_prefix. Grade the LIVE RECEIPTS: census -- generate_ai_context.py drops off entirely (6 -> 0), lsa.py goes 1 -> 2 with its remaining hits being only the default constant and a migration comment, common.py stays 1, editing_prompt.txt stays 2; counts 0 0 0 0 -> 4 3 3 1 for blogs.nix, lsa.py, generate_ai_context.py, common.py; jq MISSING x4 -> futureproof x4, and this is the ONLY probe that witnesses the nixos-rebuild ignition, because the Python helper reads through DEFAULT_PERMALINK_PREFIX and prints the same four lines either way; helper probe AttributeError -> four "N /futureproof/demo" lines; get('url' 1 -> 0; posts --fmt dated-slugs byte-identical modulo the cold-cache stderr line, which proves no regression and NOT the new path, since every real post carries a frontmatter permalink and never reaches the fallback. If the census names a file that was not in the previous context, report it and patch nothing. Then land the next step, with _config.yml now in context for both targets. (a) Template the folder into editing_prompt.txt: articleizer.py already resolves target_config and already does one placeholder substitution ([INSERT BOOK SPINE]) before the API call, so add [INSERT PERMALINK PREFIX] the same way and stop instructing every blog to write "/futureproof/[slug]/" into its frontmatter. Two sites in that file need it: the permalink construction rule and the promotional-tweet URL example. (b) Rule on generate_post_nav.py's fallback, citing _config.yml's permalink setting rather than guessing: it returns "//" where lsa.default_permalink returns "//", and one of them is wrong about what Jekyll actually serves for a permalink-less post. If _config.yml turns out not to declare permalink at all, say so and name Jekyll's default rather than inventing one. Also standing, not for this turn: three more hardwired blog specifics are visible in the existing context and want their own car later -- build_knowledge_graph.py hardcodes base_url="https://mikelev.in" as a default parameter in project_llms_txt and project_xml_sitemaps and both are called without the argument, plus a root node titled "MikeLev.in" with a hardcoded blurb; generate_hubs.py hardcodes the same base_url in its JSON-LD breadcrumb builder; gsc_historical_fetch.py hardcodes SITE_URL = "sc-domain:mikelev.in". Each is the same disease as permalink_prefix and each needs a decision about which key it reads, not just where the string moves. **5: Deliverables**: None this turn. **Opus 5 Extra**: Context verified. ## Grading the receipts Six of seven predictions landed. The seventh was a bad instrument, not a bad patch. **Census** — `generate_ai_context.py` is gone from the list entirely (6 → 0). `lsa.py` reads 2, and both survivors are the ones I named: the migration comment and `DEFAULT_PERMALINK_PREFIX`. `common.py` holds at 1, `editing_prompt.txt` at 2. **Counts** — `4 3 3 1`, exact. **jq** — `futureproof` ×4, which is the one probe that witnesses `nixos-rebuild switch`, and it fired. **Helper** — `AttributeError` → four `/futureproof/demo` lines. **posts** — byte-identical, which proves no regression and nothing more. **`grep -c "get('url'"` read 1, not 0. The prediction was wrong and the patch was right.** The patch does not *remove* the `url` spelling; it *demotes* it — `base_url or url or default` — deliberately mirroring `get_base_url`'s tolerance. A probe anchored on a string the patch preserves by design can only ever count the survivor. This is the sibling of THE EPITAPH COUNTER: there, a probe finds the patch's own comment; here, a probe finds the fallback the patch kept on purpose. The honest witness was the git diff, which is sitting in this compile's Telemetry showing the exact hunk. **I am retiring that probe rather than repairing it** — a probe that reads 1 before and 1 after teaches nothing and trains you to skip probe output. **Six files carry `futureproof` and were not in the previous context.** Per my own instruction, reported and not patched: | File | Count | Read | |---|---|---| | `foo_files.py` | 10 | router comments and URLs — narrative, not mechanism | | `assets/axis_ledger.jsonl` | 1 | append-only history; never edit retroactively | | `assets/prompts/find404s.md` | 1 | prompt text, likely a real hardwiring | | `remotes/honeybot/queries/hub_candidates.sql` | 2 | almost certainly a `LIKE '%/futureproof/%'` filter | | `remotes/honeybot/queries/hot_404_remaps_top.sql` | 1 | same shape | | `scripts/gsc/gsc_top_movers.py` | 1 | same shape | The three at the bottom are the same disease as `permalink_prefix` and join the standing list. The SQL ones are the interesting case: they filter *logged URLs*, so they need the value at query-construction time in a repo that has no Python config reader at all. ## The `_config.yml` ruling `trimnoir/_config.yml:34` says `permalink: /:slug/`. `grimoire/_config.yml` declares no `permalink:` at all, so Jekyll uses its built-in `date` style — `/:categories/:year/:month/:day/:title:output_ext`, e.g. `/2026/09/03/three-door-boot-menu-reproducible-startup.html`. So the scoreboard for a post whose frontmatter omits `permalink`: | | trimnoir serves | grimoire serves | `generate_post_nav.article_url` guesses | `lsa.default_permalink` guesses | |---|---|---|---|---| | | `//` | `///

/.html` | `/<slug>/` | `/<prefix>/<slug>` | **`generate_post_nav.py` is right for trimnoir and wrong for grimoire. `lsa.default_permalink` is wrong for both.** But the deeper finding is that *neither should be reading `permalink_prefix` for this*, because two different questions have been wearing one word: 1. *Where do I tell the model to put new posts?* → `permalink_prefix`, in `blogs.nix`. This is a **write-side** value. 2. *Where does Jekyll serve a post that never got a frontmatter permalink?* → `_config.yml`'s `permalink:`. This is a **read-side** value, and it lives in a file `blogs.nix` does not own. Last turn's car is correct about question 1 and says nothing about question 2. Making `lsa.default_permalink` "agree" with `generate_post_nav` would be papering a third value over two files that already disagree. **I am not patching either fallback this turn, because I do not yet know whether the branch is reachable.** If every post carries its own frontmatter permalink — which is what `articleizer.py` guarantees for everything it writes — then both fallbacks are unreachable dead code and the right fix is deletion or a refusal, not a third guess. Probe 4 below decides it, and Car 3 makes the branch announce itself on every future run so the question answers itself rather than being re-litigated. ## The placeholder work, and its one real hazard Two sites in `editing_prompt.txt` need templating, and the second one hardwires *two* things — the folder and the hostname `https://mikelev.in`. Templating only the folder would hand grimoire (`http://nixos.local:4003`) a tweet example pointing at the public site. So: two placeholders. **I renamed your `[INSERT PERMALINK PREFIX]` to `[INSERT PERMALINK PATTERN]`.** The value substituted is a whole permalink shape (`/futureproof/[slug]/`), not a bare segment, because a bare segment forces the template to spell `/[PREFIX]/[slug]/` and an empty prefix then yields a doubled slash — the exact failure `default_permalink` was written to avoid. The name follows the value. Same reasoning that made `learn` become `brief`. The load-bearing detail is **ordering**. If `editing_prompt.txt` lands before `articleizer.py`, the literal string `[INSERT PERMALINK PATTERN]` reaches Gemini and gets written into a real permalink. So the reader ships first, and Car 1 carries an airlock that refuses the API call on any unsubstituted `[INSERT ...]`. That airlock has a self-referential trap I have to design around: **an article about this system quotes placeholder names in its prose — this one does.** A guard run on the assembled prompt would refuse to publish the very article documenting the mechanism. So the guard runs on the *template* before the article text is substituted in, where it structurally cannot see the article. THE INSTRUMENT BECOMES BAIT, dodged by ordering rather than by an exception list. One note on the transport: the render canary in this compile's Manifest **arrived wrapped** in markdown link syntax, so something between the compiler and me rewrote the payload. The canary sits *outside* a code fence; every byte I quote into a SEARCH block below comes from *inside* one, and the `editing_prompt.txt` body shows its `https://mikelev.in/...` URL unwrapped. That is consistent with fence contents surviving and unfenced prose being linkified, so I am treating the fenced source as trustworthy — but Car 2 block 2 quotes a URL, so you are told. ## 1. PROBES ```bash grep -c 'futureproof' scripts/articles/editing_prompt.txt grep -n 'INSERT ' scripts/articles/editing_prompt.txt .venv/bin/python -c "import sys;sys.path.insert(0,'scripts/articles');import articleizer as A;t=open('scripts/articles/editing_prompt.txt').read();print('pattern',t.count(A.PERMALINK_PATTERN_PLACEHOLDER),'baseurl',t.count(A.BASE_URL_PLACEHOLDER))" .venv/bin/python -c " import sys, os, glob sys.path.insert(0,'scripts/articles') import lsa for k in ('1','3'): d = lsa.load_targets()[k]['path'] fs = sorted(glob.glob(os.path.join(d,'*.md'))) print(k, sum(1 for f in fs if not lsa.fast_get_sort_order(f)[1]), 'of', len(fs), 'posts carry no frontmatter permalink')" grep -c 'no frontmatter permalink' scripts/articles/generate_post_nav.py ``` **Line 1**: 2 → 0. The last hardwiring in the file that *causes* the frontmatter. **Line 2**: 2 lines → 4 lines. `FULL ARTICLE` and `BOOK SPINE` before; those two plus the permalink rule and the tweet example after. **Line 3 is the binding probe, and the one that couples the two cars.** It imports `articleizer` for its *own* constant names and counts them in the template, so a typo or a drifted name fails rather than passing quietly. BEFORE it raises `AttributeError: module 'articleizer' has no attribute 'PERMALINK_PATTERN_PLACEHOLDER'` — that traceback is the reading. AFTER: `pattern 2 baseurl 1`. It also witnesses that `import lsa` resolves inside `articleizer`, which is the one new import edge this turn adds. **Line 4 is a CENSUS, not a straddle** — it reads the same before and after by design, and I am saying that up front so its unchanged output is not mistaken for a failed patch. It uses `fast_get_sort_order`, which reads only the YAML block, so ~1,450 files cost a frontmatter scan rather than a full read. Its job is to rule on whether next turn needs a fallback car at all. A `0 of N` reading on both targets means both fallbacks are unreachable and the correct next patch is a refusal, not a third guess. **Line 5**: 0 → 1, Car 3's warning string. The comment above it deliberately spells the phrase differently so the count is the print statement alone. **Behavioral honesty**: none of these executes the substitution itself. `import articleizer` runs module top-level only. The first true behavioral witness is your next real `article` or `grim` run, and there the receipt to watch for is the frontmatter permalink of the post that comes out. ## 2. NEXT CONTEXT ```text scripts/articles/articleizer.py scripts/articles/editing_prompt.txt scripts/articles/generate_post_nav.py scripts/articles/lsa.py /home/mike/repos/trimnoir/_config.yml /home/mike/repos/grimoire/_config.yml /home/mike/repos/nixos/blogs.nix scripts/articles/build_knowledge_graph.py scripts/articles/generate_hubs.py scripts/articles/gsc_historical_fetch.py scripts/gsc/gsc_top_movers.py ! grep -c 'futureproof' scripts/articles/editing_prompt.txt ! grep -n 'INSERT ' scripts/articles/editing_prompt.txt ! .venv/bin/python -c "import sys;sys.path.insert(0,'scripts/articles');import articleizer as A;t=open('scripts/articles/editing_prompt.txt').read();print('pattern',t.count(A.PERMALINK_PATTERN_PLACEHOLDER),'baseurl',t.count(A.BASE_URL_PLACEHOLDER))" ! .venv/bin/python -c " import sys, os, glob sys.path.insert(0,'scripts/articles') import lsa for k in ('1','3'): d = lsa.load_targets()[k]['path'] fs = sorted(glob.glob(os.path.join(d,'*.md'))) print(k, sum(1 for f in fs if not lsa.fast_get_sort_order(f)[1]), 'of', len(fs), 'posts carry no frontmatter permalink')" ! grep -c 'no frontmatter permalink' scripts/articles/generate_post_nav.py ``` ## 3. PATCHES Same single-line SEARCH discipline as last turn: the payload strips blank lines that the real files carry (last turn's diffs prove it), so every anchor below is one line, or a unique substring inside one line. **Car 1 — `articleizer.py` learns the folder.** Three blocks, one commit. **This car must land before Car 2.** ```text Target: scripts/articles/articleizer.py [[[SEARCH]]] import common [[[DIVIDER]]] import common import lsa [[[REPLACE]]] ``` ```text Target: scripts/articles/articleizer.py [[[SEARCH]]] SPINE_PLACEHOLDER = "[INSERT BOOK SPINE]" [[[DIVIDER]]] SPINE_PLACEHOLDER = "[INSERT BOOK SPINE]" # THE BLOG FOLDER REACHES THE MODEL (banked 2026-09-04). editing_prompt.txt is # what CAUSES every post's frontmatter to say /futureproof/, so leaving it # hardwired meant blogs.nix could declare a folder that nothing ever wrote. # A PATTERN, NOT A PREFIX, and the name follows the value: what gets # substituted is a whole permalink shape ("/futureproof/[slug]/"), because a # bare segment would force the template to spell "/[PREFIX]/[slug]/" and an # empty prefix would then yield a doubled slash -- the exact case # default_permalink exists to avoid. That same function builds this string, # so the instruction the model receives and the fallback lsa.py uses can never # disagree: one function, both answers. # BASE URL RIDES ALONG because the tweet example hardwired the HOSTNAME too, # and templating only the folder would hand grimoire the public site's host. PERMALINK_PATTERN_PLACEHOLDER = "[INSERT PERMALINK PATTERN]" BASE_URL_PLACEHOLDER = "[INSERT BASE URL]" [[[REPLACE]]] ``` ```text Target: scripts/articles/articleizer.py [[[SEARCH]]] full_prompt = prompt_template.replace(PROMPT_PLACEHOLDER, article_text) [[[DIVIDER]]] # SUBSTITUTE INTO THE TEMPLATE, BEFORE THE ARTICLE LANDS. The order is # load-bearing for the guard below: an article ABOUT this system quotes # placeholder names in its prose (the one introducing this code does), # so a guard run on the ASSEMBLED prompt would refuse to publish the # very article documenting the mechanism -- THE INSTRUMENT BECOMES # BAIT. Scanning the template alone cannot see the article and so # cannot be baited by it. prompt_prefix = lsa.permalink_prefix(target_config) # `is None`, never `or`: target 4 declares base_url = "" ON PURPOSE # (a private wiki with no public URL), and an or-chain would hand it # the public site's hostname. Mirrors this file's own canonical_url # logic further down, and permalink_prefix's None-vs-falsy rule. prompt_base_url = target_config.get("base_url") if prompt_base_url is None: prompt_base_url = target_config.get("url", "https://mikelev.in") prompt_template = prompt_template.replace( PERMALINK_PATTERN_PLACEHOLDER, lsa.default_permalink("[slug]", prompt_prefix) + "/") prompt_template = prompt_template.replace( BASE_URL_PLACEHOLDER, prompt_base_url.rstrip("/")) # THE UNSUBSTITUTED-PLACEHOLDER AIRLOCK: if editing_prompt.txt carries # a placeholder this file does not know -- a typo, or a template # patched ahead of its reader -- the literal "[INSERT ...]" reaches the # model and lands verbatim in a permalink. Refuse instead. FULL ARTICLE # and BOOK SPINE are exempt because they are substituted after this # point, which is the only reason this guard can run here at all. leftover = sorted(set(re.findall(r"\[INSERT [A-Z ]+\]", prompt_template)) - {PROMPT_PLACEHOLDER, SPINE_PLACEHOLDER}) if leftover: print(f"❌ Unsubstituted placeholder(s) in {PROMPT_FILENAME}: {leftover}") print(" Refusing to call the API; the model would copy the literal text into your frontmatter.") return full_prompt = prompt_template.replace(PROMPT_PLACEHOLDER, article_text) [[[REPLACE]]] ``` **Car 2 — `editing_prompt.txt` stops naming a folder it cannot know.** Two blocks, one commit. Both SEARCHes are unique substrings chosen to avoid the line's leading indentation and its trailing double-space markdown break, neither of which survives this payload reliably. ```text Target: scripts/articles/editing_prompt.txt [[[SEARCH]]] use it to construct the permalink string in the format /futureproof/[slug]/. [[[DIVIDER]]] use it to construct the permalink string in the format [INSERT PERMALINK PATTERN] -- copy that shape verbatim, keeping its leading and trailing slashes, and put your slug where [slug] appears. [[[REPLACE]]] ``` ```text Target: scripts/articles/editing_prompt.txt [[[SEARCH]]] (e.g., https://mikelev.in/futureproof/[slug]/) [[[DIVIDER]]] (e.g., [INSERT BASE URL][INSERT PERMALINK PATTERN]) [[[REPLACE]]] ``` **Car 3 — the fallback announces itself.** One block. This does not pick a winner between the two disagreeing fallbacks; it makes the branch report whether it ever fires, converting a debate into a receipt. ```text Target: scripts/articles/generate_post_nav.py [[[SEARCH]]] nav = build_nav(metadata) [[[DIVIDER]]] # THE FALLBACK'S REACHABILITY, SELF-REPORTING (2026-09-04). article_url # above invents "/<slug>/" for a post with no frontmatter permalink, while # lsa.default_permalink invents "/<prefix>/<slug>" for that same post. # They disagree, and NEITHER is authoritative: what Jekyll serves comes # from _config.yml's permalink: setting -- "/:slug/" for trimnoir, and the # unset `date` default for grimoire, which is neither of the two guesses. # permalink_prefix cannot settle it, because that value answers a # different question (where articleizer WRITES new posts), and it lives in # a file that does not own Jekyll's routing. # So: do not pick a winner blind. Make the branch speak. Silence means # every post carried its own permalink and the disagreement never fired, # which is the reading that decides whether a fix is owed at all. unlinked = [i["filename"] for i in metadata if not str(i.get("permalink") or "").strip()] if unlinked: print(f"⚠️ {len(unlinked)} post(s) have no frontmatter permalink; their nav keys are GUESSED as /<slug>/.") for name in unlinked[:5]: print(f" - {name}") if len(unlinked) > 5: print(f" ... and {len(unlinked) - 5} more") nav = build_nav(metadata) [[[REPLACE]]] ``` **Ignition: none required.** All three files are plain Python or data read at call time; probe 3 imports the patched `articleizer` and reads the patched template on the same line, and probe 5 greps the patched file. Nothing here is a shellHook, a daemon, or a cached artifact. **Choreography** — all three cars are inside the pipulate repo this time, so the ordinary loop applies, with the one ordering constraint: ```text patch, app, d, m # Car 1 -- articleizer.py, the READER, must go first patch, app, d, m # Car 2 -- editing_prompt.txt, the TEMPLATE patch, app, d, m # Car 3 -- generate_post_nav.py blast ``` On your remark about splitting the article by cars: that is what they already are. A car is defined as one commit story, so the car boundary and the section boundary are the same line — which is why `m` produces a usable heading for each one for free. ## 4. PROMPT ```text Three cars templated the blog folder into the prompt that writes it. Grade the LIVE RECEIPTS: editing_prompt.txt futureproof 2 -> 0; 'INSERT ' 2 lines -> 4 lines; the binding probe AttributeError -> "pattern 2 baseurl 1", which witnesses BOTH that the template carries the placeholders and that articleizer's own constant names match them; generate_post_nav 'no frontmatter permalink' 0 -> 1. The permalink census is a CENSUS, not a straddle -- it reads identically before and after by design, and its number is the ruling, not a receipt of this patch. Then rule on the fallback, using that census number. If it reads "0 of N" for both targets, both fallbacks are unreachable dead code: emit the car that replaces lsa.default_permalink's guess and generate_post_nav's guess with a single shared refusal path (warn loudly, and do not invent a URL that Jekyll does not serve), and say plainly that the change is unwitnessable today because nothing reaches it. If it reads nonzero for either target, name the affected posts and emit instead a reader for _config.yml's permalink: setting -- that file, not blogs.nix, is what decides where Jekyll serves a permalink-less post, and grimoire declares no permalink: at all, so its answer is Jekyll's built-in date style and not /<slug>/. Also confirm, from the first real `article` or `grim` run after these cars: the frontmatter permalink of the post that comes out, and that no literal "[INSERT " string survived into it. That is the only behavioral witness these cars have; every probe above is structural. Standing, still unpatched, now nine files across two categories. Python that hardcodes the hostname: build_knowledge_graph.py (base_url="https://mikelev.in" as a default parameter in project_llms_txt and project_xml_sitemaps, both called without the argument, plus a root node titled "MikeLev.in" with a hardcoded blurb); generate_hubs.py (same base_url in its JSON-LD breadcrumb builder); gsc_historical_fetch.py (SITE_URL = "sc-domain:mikelev.in"). And the census surfaced four more that carry the FOLDER outside any Python config reader: scripts/gsc/gsc_top_movers.py, assets/prompts/find404s.md, and remotes/honeybot/queries/hub_candidates.sql plus hot_404_remaps_top.sql. The SQL pair is the interesting one and needs its own decision: those queries filter logged URLs and run in a repo with no blogs.json reader at all, so "move the string" is not available -- either they are generated with the value injected, or they stay hardwired and say so. foo_files.py's ten hits are narrative comments and assets/axis_ledger.jsonl is append-only history; neither is a hardwiring and neither should be touched. ``` ## 5. EXTERNAL DELIVERABLES None outside the repo. One unrelated one-liner if the recurring Ruff diagnostic bothers you — `generate_llms_txt.py` has no trailing newline, so every compile prints `Found 1 error`, which is the shape that trains you to skip Ruff output. Fix it in the shell rather than through a patch, since a trailing newline is not a thing SEARCH/REPLACE expresses reliably: ```bash printf '\n' >> scripts/articles/generate_llms_txt.py && ruff check scripts/articles/generate_llms_txt.py ``` **MikeLev.in**: > Same commands, run twice, one change between them. Where the readings > differ is what the change did; the diff in the middle is the receipt. **1: Probe** (BEFORE: hand-run, nothing changed yet): ```bash Blast Radius Check to establish bisection Left-hand Causal Boundary. It is a Popper-thing. Science. On branch main Your branch is up to date with 'origin/main'. nothing to commit, working tree clean (nix) pipulate $ grep -c 'futureproof' scripts/articles/editing_prompt.txt grep -n 'INSERT ' scripts/articles/editing_prompt.txt .venv/bin/python -c "import sys;sys.path.insert(0,'scripts/articles');import articleizer as A;t=open('scripts/articles/editing_prompt.txt').read();print('pattern',t.count(A.PERMALINK_PATTERN_PLACEHOLDER),'baseurl',t.count(A.BASE_URL_PLACEHOLDER))" .venv/bin/python -c " import sys, os, glob sys.path.insert(0,'scripts/articles') import lsa for k in ('1','3'): d = lsa.load_targets()[k]['path'] fs = sorted(glob.glob(os.path.join(d,'*.md'))) print(k, sum(1 for f in fs if not lsa.fast_get_sort_order(f)[1]), 'of', len(fs), 'posts carry no frontmatter permalink')" grep -c 'no frontmatter permalink' scripts/articles/generate_post_nav.py 2 19:[INSERT BOOK SPINE] 25:[INSERT FULL ARTICLE] Traceback (most recent call last): File "<string>", line 1, in <module> AttributeError: module 'articleizer' has no attribute 'PERMALINK_PATTERN_PLACEHOLDER' 1 0 of 1446 posts carry no frontmatter permalink 3 0 of 126 posts carry no frontmatter permalink 0 (nix) pipulate $ ``` **2: Context** (AFTER: the same probes re-run by the compiler as `!` lines): ```text # adhoc.txt _ _ _ to set context____ _ _ ___ ____ _ Simpson Couch Gag Here (explain anything to the audience you feel needs it explained)G # / \ __| | | | | | ___ ___ / ___| | | |/ _ \| _ \| | # ahe/ _ \ / _` | | |_| |/ _ \ / __| | | | |_| | | | | |_) | | Making the blogging system less hardwired to a particular subfolder. # ahc ___ \ (_| | | _ | (_) | (__ | |___| _ | |_| | __/|_| Getting there. # /_/ \_\__,_| |_| |_|\___/ \___| \____|_| |_|\___/|_| (_) # Ad Hoc CHOP: The Not-Managed-by-Git Safe-for-Client-Data place # OPTIONAL BUT BIG FOR FULL CONTEXT-WINDOW STORYTELLING # ! python scripts/articles/lsa.py -t 1 --reverse --fmt dated-slugs # <-- The "Rolling Pin" that gives the 40K foot book-spine view of book-ore. # GLOSSARY.md # <-- Like the back of a J.R.R. Tolkien book but always growing in size as `prompt_foo.py` gets scars and shrinks. # scripts/articles/lsa.py # <-- 2nd Brain query-engine for `rgx`, `rgxc` & `posts` Jekyll-inspired Memory Externalization for Hackers. # ~/repos/nixos/autognome.py # <-- Letting the AIs really understand my environment (The Brave Little Tailor punches above Their Weight Class proving the dunning-kruger effect the gate-keeper's (lower-case) lament.) init.lua # <-- Daily driver hot-keys that overlap with aliases in flake.nix. `<leader>m` makes it Science (this process)! # STILL BIG BUT LESS OPTIONAL (especially flake.nix) flake.nix # <-- THE ONE BIG THING TO INCLUDE Infrastructure as Code (IaC) tells LLM about your system down to the metal # prompt_foo.py # <-- This very content-compiling system # foo_files.py # <-- This is the router, evolving book outline and the things you pin-up to produced the recursive self-improvement loops # TINY ILLUMINATING (OK to include every time / automatically = `apply.py`, `.gitignore`, `.gitattributes`) # requirements.in # <-- All known dependencies and (necessary) version pinning. WORA gotcha's exposed. # __init__.py # <-- Master versioning # pyproject.toml # <-- The PyPI Packaging details # OPTIONAL ACTUATORS (cheap and good to include to expand the AI's capabilities) # cli.py # <-- Catch-all actuator for PyPI envs, Python anchoring, MCP tool-call (plus alternatives) and **kwargs like wrapping for CLI # scripts/xp.py # <-- Transforms host OS copy-paste buffer player-piano music into context-payload. # scripts/ai.py # <-- How I constantly use local AI to write git commit messages with `m` alias. # scripts/crawl.py # <-- Feel free to ask for something to be crawled and included in the next turn. # scripts/weblogin.py # <-- Lets the user "warm up" the cache for their web logins at their leisure on a profile that persists. # MISCELLANEOUS (rare to include but sometimes critical) # scripts/foo_cartridge.py # Needs description # scripts/foo_replay.py # Needs description # release.py # <-- How everything ends up where it does (GitHub, PyPI, etc.) # imports/voice_synthesis.py # <-- The wand can talk to you # imports/ascii_displays.py # <-- Where all the ASCII Art lives # scripts/release/version_sync.py # <-- Needs to be wrapped into release.py and eliminated, I think. # --- Under this line is were you paste what the AI gives you --- # --- We call it context but it's really just the right-hand --- # --- blast-radius of the "probes" to make this all science. --- # Carry-over as the important work-in-progress parts of the project here just # like above but not as long-standing overarching to the framework but rather # for the current hot spots actively being worked on. # STICKBUG & MOTHER CAT KATA (WORKING ON THE CHAPTER) # assets/installer/mck.sh # assets/installer/replay.sh # scripts/walk.py # scripts/walk_cartridge.py # scripts/walk_compile.py # assets/trails/first_context.yaml # assets/trails/practice.yaml # assets/trails/public_walk.yaml # assets/trails/botify_pageworkers.yaml # scripts/connectors/README.md # scripts/connectors/botify.py # scripts/connectors/confluence.py # scripts/connectors/gmail.py # scripts/connectors/gsc.py # scripts/connectors/jira.py # scripts/connectors/mcp.py # scripts/connectors/mcp_warm.py # scripts/connectors/sheets.py # scripts/connectors/slack.py # scripts/connectors/wallet.py # scripts/bookmark_import.py # scripts/boot_menu.py # scripts/mother_cat.py # scripts/sources_menu.py # scripts/weblogin.py # tools/scraper_tools.py # --- START THIS DISCUSSION --- # Context 1 (Edit-in selections from above and add new files immediately below) # # /home/mike/repos/grimoire/_posts/2026-09-03-three-door-boot-menu-reproducible-startup.md # /home/mike/repos/trimnoir/_posts/2026-09-04-closing-the-loop-replayable-ai-workflows-and-manual-edits.md # # # CORE SLIDESHOW FILES # nixops.sh # <-- You've heard of GitOPs? Well, this is NixOPs. # remotes/honeybot/hooks/post-receive # <-- Ever hear of GitHub Pages? Or github.io? This is that. # remotes/honeybot/nixos/configuration.nix # <-- It's as if Pipulate had kids. Spy kids. # remotes/honeybot/scripts/stream.py # <-- Starts the TV Channel streaming to YouTube-live via OBS from Nginx Honeybot XFCE Desktop. Clear? # remotes/honeybot/scripts/score.py # <-- Where "Greetings Entity" slideshow reads on post-receive interrupts # remotes/honeybot/scripts/card.py # <-- Just added for station identification breaks # remotes/honeybot/scripts/forest.py # <-- Likewise, just added for the new storytelling system on Honeybot # remotes/honeybot/scripts/test_forest.py # <-- Test Honeybot station identification sequence on Pipulate Prime # remotes/honeybot/scripts/logs.py # <-- The TV Show is mostly Nginx `access.log` files tailed and piped through Python to colorize (this). # remotes/honeybot/scripts/content_loader.py # <-- Tricky TV programming & scheduling stuff. Absolute versus relative timing. Loops. Interrupts. # remotes/honeybot/scripts/db.py # <-- But you can't keep your weblogs forever! And we want trending. And data-mining. Here's how. # imports/voice_synthesis.py # <-- The wand can talk to you (not sure if I'm keeping it in Honeybot chapter) # # # JEKYLL PUBLISHING # ~/.config/pipulate/blogs.json # <-- CAUTION! Derived from ~/repos/nixos/blogs.nix # scripts/articles/publishizer.py # <-- Orchestrates different publishing workflows per target blog. # scripts/articles/common.py # <-- Self-explanatory # scripts/articles/articleizer.py # <-- Transforms raw article.txt to formal Jekyll markdown format # scripts/articles/editing_prompt.txt # <-- Forcing response into strict JSON data structure # scripts/articles/sanitizer.py # <-- Scrubs PII # scripts/articles/gsc_historical_fetch.py # scripts/articles/contextualizer.py # <-- Builds JSON summaries of articles in `_posts/context/` called "Holographic Shards". # scripts/articles/confluenceizer.py # <-- Idempotent Jekyll-to-Confluence corporate wiki # scripts/articles/googledocizer.py # <-- Just added # scripts/articles/build_knowledge_graph.py # <-- Topically load-balances site using hierarchical K-Means keyword clustering groups # scripts/articles/generate_ai_context.py # <-- AIs WILL interrogate your repo. This gives epic context of article URLs for drill-down. # scripts/articles/generate_hubs.py # <-- Uses just-produced link-graph data to generate each of the new hubs it suggests # scripts/articles/generate_llms_txt.py # <-- Builds an llms.txt based on the auto-organized structure suggested here # scripts/articles/generate_redirects.py # <-- Generates redirect map above hub-churn suggests is needed # scripts/articles/sanitize_redirects.py # <-- Deals with follow-up meticulous pedantic detail required for a good Nginx redirect map # # /home/mike/repos/trimnoir/_layouts/default.html # /home/mike/repos/trimnoir/_layouts/forward.html # /home/mike/repos/trimnoir/_layouts/home.html # /home/mike/repos/trimnoir/_layouts/page.html # /home/mike/repos/trimnoir/_layouts/plain.html # /home/mike/repos/trimnoir/_layouts/post.html # /home/mike/repos/grimoire/flake.nix # /home/mike/repos/grimoire/_layouts/default.html # /home/mike/repos/grimoire/index.md # # /home/mike/repos/nixos/blogs.nix # /home/mike/.config/pipulate/blogs.json # scripts/articles/lsa.py # scripts/articles/link_injector.py # scripts/articles/generate_post_nav.py # Context 2 # /home/mike/repos/nixos/blogs.nix # /home/mike/.config/pipulate/blogs.json # scripts/articles/lsa.py # scripts/articles/generate_ai_context.py # scripts/articles/common.py # scripts/articles/generate_llms_txt.py # scripts/articles/generate_post_nav.py # scripts/articles/articleizer.py # scripts/articles/editing_prompt.txt # /home/mike/repos/trimnoir/_config.yml # /home/mike/repos/grimoire/_config.yml # ! rg -c 'futureproof' --glob '!AI_CONTEXT.md' . | sort # ! for f in "$HOME/repos/nixos/blogs.nix" scripts/articles/lsa.py scripts/articles/generate_ai_context.py scripts/articles/common.py; do printf '%3s %s\n' "$(grep -c permalink_prefix "$f")" "$f"; done # ! jq -r 'to_entries[] | "\(.key) permalink_prefix=\(.value.permalink_prefix // "MISSING")"' "$HOME/.config/pipulate/blogs.json" # ! .venv/bin/python -c "import sys; sys.path.insert(0,'scripts/articles'); import lsa; [print(k, lsa.default_permalink('demo', lsa.permalink_prefix(v))) for k, v in sorted(lsa.load_targets().items())]" # ! grep -c "get('url'" scripts/articles/generate_llms_txt.py # ! grep -n permalink "$HOME/repos/trimnoir/_config.yml" "$HOME/repos/grimoire/_config.yml" 2>&1 | head -10 # ! posts -t 1 --fmt dated-slugs --last 3 # Context 3 scripts/articles/articleizer.py scripts/articles/editing_prompt.txt scripts/articles/generate_post_nav.py scripts/articles/lsa.py /home/mike/repos/trimnoir/_config.yml /home/mike/repos/grimoire/_config.yml /home/mike/repos/nixos/blogs.nix scripts/articles/build_knowledge_graph.py scripts/articles/generate_hubs.py scripts/articles/gsc_historical_fetch.py scripts/gsc/gsc_top_movers.py ! grep -c 'futureproof' scripts/articles/editing_prompt.txt ! grep -n 'INSERT ' scripts/articles/editing_prompt.txt ! .venv/bin/python -c "import sys;sys.path.insert(0,'scripts/articles');import articleizer as A;t=open('scripts/articles/editing_prompt.txt').read();print('pattern',t.count(A.PERMALINK_PATTERN_PLACEHOLDER),'baseurl',t.count(A.BASE_URL_PLACEHOLDER))" ! .venv/bin/python -c " import sys, os, glob sys.path.insert(0,'scripts/articles') import lsa for k in ('1','3'): d = lsa.load_targets()[k]['path'] fs = sorted(glob.glob(os.path.join(d,'*.md'))) print(k, sum(1 for f in fs if not lsa.fast_get_sort_order(f)[1]), 'of', len(fs), 'posts carry no frontmatter permalink')" ! grep -c 'no frontmatter permalink' scripts/articles/generate_post_nav.py # --- END `adhoc.txt` TEMPLATE --- ``` **3: Patches** (the one change between the readings): I'm doing all the so-called Cars at once after all. ```diff Blast Radius Check to establish bisection Left-hand Causal Boundary. It is a Popper-thing. Science. On branch main Your branch is up to date with 'origin/main'. nothing to commit, working tree clean (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/articleizer.py'. (nix) pipulate $ d diff --git a/scripts/articles/articleizer.py b/scripts/articles/articleizer.py index b191a2a2..c77f7935 100644 --- a/scripts/articles/articleizer.py +++ b/scripts/articles/articleizer.py @@ -9,6 +9,7 @@ import llm import argparse import time import common +import lsa # --- CONFIGURATION --- CONFIG_DIR = Path.home() / ".config" / "articleizer" (nix) pipulate $ m 📝 Committing: chore: Import lsa module [main c46beaa3] chore: Import lsa module 1 file changed, 1 insertion(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/articleizer.py'. (nix) pipulate $ d diff --git a/scripts/articles/articleizer.py b/scripts/articles/articleizer.py index c77f7935..10642ced 100644 --- a/scripts/articles/articleizer.py +++ b/scripts/articles/articleizer.py @@ -24,6 +24,20 @@ DEFAULT_MODEL = 'gemini-flash-latest' # DEFAULT_MODEL = 'gemini-flash-lite-latest' SPINE_PLACEHOLDER = "[INSERT BOOK SPINE]" +# THE BLOG FOLDER REACHES THE MODEL (banked 2026-09-04). editing_prompt.txt is +# what CAUSES every post's frontmatter to say /futureproof/, so leaving it +# hardwired meant blogs.nix could declare a folder that nothing ever wrote. +# A PATTERN, NOT A PREFIX, and the name follows the value: what gets +# substituted is a whole permalink shape ("/futureproof/[slug]/"), because a +# bare segment would force the template to spell "/[PREFIX]/[slug]/" and an +# empty prefix would then yield a doubled slash -- the exact case +# default_permalink exists to avoid. That same function builds this string, +# so the instruction the model receives and the fallback lsa.py uses can never +# disagree: one function, both answers. +# BASE URL RIDES ALONG because the tweet example hardwired the HOSTNAME too, +# and templating only the folder would hand grimoire the public site's host. +PERMALINK_PATTERN_PLACEHOLDER = "[INSERT PERMALINK PATTERN]" +BASE_URL_PLACEHOLDER = "[INSERT BASE URL]" def scan_corpus(output_dir): (nix) pipulate $ m 📝 Committing: chore: Update placeholders in articleizer.py [main 8d8ab169] chore: Update placeholders in articleizer.py 1 file changed, 14 insertions(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/articleizer.py'. (nix) pipulate $ d diff --git a/scripts/articles/articleizer.py b/scripts/articles/articleizer.py index 10642ced..5cd9b581 100644 --- a/scripts/articles/articleizer.py +++ b/scripts/articles/articleizer.py @@ -395,6 +395,38 @@ def main(): with open(PROMPT_FILENAME, 'r', encoding='utf-8') as f: prompt_template = f.read() + # SUBSTITUTE INTO THE TEMPLATE, BEFORE THE ARTICLE LANDS. The order is + # load-bearing for the guard below: an article ABOUT this system quotes + # placeholder names in its prose (the one introducing this code does), + # so a guard run on the ASSEMBLED prompt would refuse to publish the + # very article documenting the mechanism -- THE INSTRUMENT BECOMES + # BAIT. Scanning the template alone cannot see the article and so + # cannot be baited by it. + prompt_prefix = lsa.permalink_prefix(target_config) + # `is None`, never `or`: target 4 declares base_url = "" ON PURPOSE + # (a private wiki with no public URL), and an or-chain would hand it + # the public site's hostname. Mirrors this file's own canonical_url + # logic further down, and permalink_prefix's None-vs-falsy rule. + prompt_base_url = target_config.get("base_url") + if prompt_base_url is None: + prompt_base_url = target_config.get("url", "https://mikelev.in") + prompt_template = prompt_template.replace( + PERMALINK_PATTERN_PLACEHOLDER, + lsa.default_permalink("[slug]", prompt_prefix) + "/") + prompt_template = prompt_template.replace( + BASE_URL_PLACEHOLDER, prompt_base_url.rstrip("/")) + # THE UNSUBSTITUTED-PLACEHOLDER AIRLOCK: if editing_prompt.txt carries + # a placeholder this file does not know -- a typo, or a template + # patched ahead of its reader -- the literal "[INSERT ...]" reaches the + # model and lands verbatim in a permalink. Refuse instead. FULL ARTICLE + # and BOOK SPINE are exempt because they are substituted after this + # point, which is the only reason this guard can run here at all. + leftover = sorted(set(re.findall(r"\[INSERT [A-Z ]+\]", prompt_template)) + - {PROMPT_PLACEHOLDER, SPINE_PLACEHOLDER}) + if leftover: + print(f"❌ Unsubstituted placeholder(s) in {PROMPT_FILENAME}: {leftover}") + print(" Refusing to call the API; the model would copy the literal text into your frontmatter.") + return full_prompt = prompt_template.replace(PROMPT_PLACEHOLDER, article_text) # --- BOOK SPINE INJECTION (40K-foot view for the editing model) --- (nix) pipulate $ m 📝 Committing: chore: Prevent placeholder leaks in articleizer template [main 14e35628] chore: Prevent placeholder leaks in articleizer template 1 file changed, 32 insertions(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/editing_prompt.txt'. (nix) pipulate $ d diff --git a/scripts/articles/editing_prompt.txt b/scripts/articles/editing_prompt.txt index 0f92e40b..f788cd7c 100644 --- a/scripts/articles/editing_prompt.txt +++ b/scripts/articles/editing_prompt.txt @@ -72,7 +72,7 @@ The JSON object must conform to the following schema: * **yaml_updates**: * title: Populate with your **preferred title** from your analysis. * description: Populate with the meta_description you craft. This should be 100-160 characters. - * permalink: Generate the SEO-optimized filename slug (e.g., primary-keywords.md) and use it to construct the permalink string in the format /futureproof/[slug]/. + * permalink: Generate the SEO-optimized filename slug (e.g., primary-keywords.md) and use it to construct the permalink string in the format [INSERT PERMALINK PATTERN] -- copy that shape verbatim, keeping its leading and trailing slashes, and put your slug where [slug] appears. * keywords: Populate with the comma-separated meta_keywords string. * **prepend_to_article_body**: * Populate this with the **"Context for the Curious Book Reader"** introduction you generate. (nix) pipulate $ m 📝 Committing: chore: Update editing prompt with permalink instruction [main 60f9e99a] chore: Update editing prompt with permalink instruction 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/editing_prompt.txt'. (nix) pipulate $ d diff --git a/scripts/articles/editing_prompt.txt b/scripts/articles/editing_prompt.txt index f788cd7c..7d3082e1 100644 --- a/scripts/articles/editing_prompt.txt +++ b/scripts/articles/editing_prompt.txt @@ -85,7 +85,7 @@ The JSON object must conform to the following schema: * VALIDATION: Ensure the snippet is followed by a double line-break (blank line) in the source text. Do NOT select words from the middle of a sentence or paragraph. * **book_analysis_content**: This top-level key contains all the valuable analysis that doesn't get edited into the file directly but is used for the book-writing process. * authors_imprint: Populate with the first-person summary. - * promotional_tweet: Write a tweet for X.com where we can announce this article so that it will most effectively reach the target audience whom it will most help. Use the same reasoning you used for choosing the top-choice SEO title and URL and adapt it for a tweet. Include the naked URL (e.g., https://mikelev.in/futureproof/[slug]/) and NOT markdown format because of how X.com works. Include the 2 or 3 best hashtags, space allowing. Target the size of a traditional "non-verified" tweet (max 280 characters). + * promotional_tweet: Write a tweet for X.com where we can announce this article so that it will most effectively reach the target audience whom it will most help. Use the same reasoning you used for choosing the top-choice SEO title and URL and adapt it for a tweet. Include the naked URL (e.g., [INSERT BASE URL][INSERT PERMALINK PATTERN]) and NOT markdown format because of how X.com works. Include the 2 or 3 best hashtags, space allowing. Target the size of a traditional "non-verified" tweet (max 280 characters). * title_brainstorm: Populate with the full list of your 3-5 title ideas, including rationales and filenames. Ensure the first entry is your preferred choice. Do not prepend the date on the filename (that happens automatically). * content_potential_and_polish: Structure the "Core Strengths" and "Constructive Polish" suggestions here. * ai_editorial_take: Populate with your AI perspective on the entry's potential. What surprised you most? Don't use the exact same thing as the previous AI participating in the dialogue. Find some absolutely new angle or perspective that isn't mentioned yet in the article. (nix) pipulate $ m 📝 Committing: chore: Update editing\_prompt.txt with URL and hashtag guidance [main 1e94e333] chore: Update editing\_prompt.txt with URL and hashtag guidance 1 file changed, 1 insertion(+), 1 deletion(-) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'scripts/articles/generate_post_nav.py'. (nix) pipulate $ d diff --git a/scripts/articles/generate_post_nav.py b/scripts/articles/generate_post_nav.py index 09cc0672..086a09e4 100644 --- a/scripts/articles/generate_post_nav.py +++ b/scripts/articles/generate_post_nav.py @@ -78,6 +78,25 @@ def main(): print(f"❌ No articles found under {posts_dir}", file=sys.stderr) sys.exit(1) + # THE FALLBACK'S REACHABILITY, SELF-REPORTING (2026-09-04). article_url + # above invents "/<slug>/" for a post with no frontmatter permalink, while + # lsa.default_permalink invents "/<prefix>/<slug>" for that same post. + # They disagree, and NEITHER is authoritative: what Jekyll serves comes + # from _config.yml's permalink: setting -- "/:slug/" for trimnoir, and the + # unset `date` default for grimoire, which is neither of the two guesses. + # permalink_prefix cannot settle it, because that value answers a + # different question (where articleizer WRITES new posts), and it lives in + # a file that does not own Jekyll's routing. + # So: do not pick a winner blind. Make the branch speak. Silence means + # every post carried its own permalink and the disagreement never fired, + # which is the reading that decides whether a fix is owed at all. + unlinked = [i["filename"] for i in metadata if not str(i.get("permalink") or "").strip()] + if unlinked: + print(f"⚠️ {len(unlinked)} post(s) have no frontmatter permalink; their nav keys are GUESSED as /<slug>/.") + for name in unlinked[:5]: + print(f" - {name}") + if len(unlinked) > 5: + print(f" ... and {len(unlinked) - 5} more") nav = build_nav(metadata) if len(nav) != len(metadata): print(f"⚠️ URL collision: {len(metadata)} posts collapsed to {len(nav)} nav keys.") (nix) pipulate $ m 📝 Committing: chore: Handle missing post permalinks in nav generation [main e6c3bd46] chore: Handle missing post permalinks in nav generation 1 file changed, 19 insertions(+) (nix) pipulate $ git push Enumerating objects: 36, done. Counting objects: 100% (36/36), done. Delta compression using up to 48 threads Compressing objects: 100% (30/30), done. Writing objects: 100% (30/30), 4.79 KiB | 2.39 MiB/s, done. Total 30 (delta 24), reused 0 (delta 0), pack-reused 0 (from 0) remote: Resolving deltas: 100% (24/24), completed with 6 local objects. To github.com:pipulate/pipulate.git d189f097..e6c3bd46 main -> main (nix) pipulate $ ``` **4: Prompt** Three cars templated the blog folder into the prompt that writes it. Grade the LIVE RECEIPTS: editing_prompt.txt futureproof 2 -> 0; 'INSERT ' 2 lines -> 4 lines; the binding probe AttributeError -> "pattern 2 baseurl 1", which witnesses BOTH that the template carries the placeholders and that articleizer's own constant names match them; generate_post_nav 'no frontmatter permalink' 0 -> 1. The permalink census is a CENSUS, not a straddle -- it reads identically before and after by design, and its number is the ruling, not a receipt of this patch. Then rule on the fallback, using that census number. If it reads "0 of N" for both targets, both fallbacks are unreachable dead code: emit the car that replaces lsa.default_permalink's guess and generate_post_nav's guess with a single shared refusal path (warn loudly, and do not invent a URL that Jekyll does not serve), and say plainly that the change is unwitnessable today because nothing reaches it. If it reads nonzero for either target, name the affected posts and emit instead a reader for _config.yml's permalink: setting -- that file, not blogs.nix, is what decides where Jekyll serves a permalink-less post, and grimoire declares no permalink: at all, so its answer is Jekyll's built-in date style and not /<slug>/. Also confirm, from the first real `article` or `grim` run after these cars: the frontmatter permalink of the post that comes out, and that no literal "[INSERT " string survived into it. That is the only behavioral witness these cars have; every probe above is structural. Standing, still unpatched, now nine files across two categories. Python that hardcodes the hostname: build_knowledge_graph.py (base_url="https://mikelev.in" as a default parameter in project_llms_txt and project_xml_sitemaps, both called without the argument, plus a root node titled "MikeLev.in" with a hardcoded blurb); generate_hubs.py (same base_url in its JSON-LD breadcrumb builder); gsc_historical_fetch.py (SITE_URL = "sc-domain:mikelev.in"). And the census surfaced four more that carry the FOLDER outside any Python config reader: scripts/gsc/gsc_top_movers.py, assets/prompts/find404s.md, and remotes/honeybot/queries/hub_candidates.sql plus hot_404_remaps_top.sql. The SQL pair is the interesting one and needs its own decision: those queries filter logged URLs and run in a repo with no blogs.json reader at all, so "move the string" is not available -- either they are generated with the value injected, or they stay hardwired and say so. foo_files.py's ten hits are narrative comments and assets/axis_ledger.jsonl is append-only history; neither is a hardwiring and neither should be touched. **5: Deliverables**: None this turn outside the repo. ```bash (nix) pipulate $ printf '\n' >> scripts/articles/generate_llms_txt.py && ruff check scripts/articles/generate_llms_txt.py All checks passed! (nix) pipulate $ ``` Are we done? Hop off the ride. This ride's stated goal is reached — dismount. This is the NOTARY BEAT: the ride ends here, is witnessed here, and is sealed here. Answer all seven beats, briefly: 0. **TL;DR**: a short, dry, neutral abstract for the TOP of the published article — written for an unfamiliar reader or AI summarizer who has never seen this system. No hype, no insider handles unexplained. 1. VERIFY: restate the goal from the top of this article and confirm (or deny) it was met, citing THIS compile's receipts, not memory. Name any ignition this ride required that never fired -- an AFTER tap taken without one is a stale BEFORE wearing the AFTER's label. 2. BANK: name everything that graduates — rule, earmark, todo, pin — as SEARCH/REPLACE patch cars against raw source present in THIS context (Target line inside the fence, one car per commit story), deletions included. A 'paste-ready' line with no Target and no SEARCH anchor is not banked; it is a hand edit the operator will not make. These BANK cars are the ONLY patches a dismount emits. 3. DANGLING: what carries forward unbanked? One line each, no essays. 4. SEED: the adhoc.txt lines (and TODO_SLUGS if narrative context is needed) for the next ride's first compile. 5. CLOSING: a closing summary for the BOTTOM of the article — the final take-away, tied to the book's larger arc where it fits naturally, never forced. Storytelling over inventory. 6. NOTARIZE: the artifact of record is the cartridge THIS compile writes after this payload is assembled, so its name is not in your context and you must not guess one. Attest the FINGERPRINT instead: this Manifest's wc -c line, its receipt count, and the newest receipt's date -- pure functions of the bytes the cartridge seals. Close with one line the operator completes after the compile: Deed: foo-________.zip (ls -t foo-*.zip | head -1) The seal is the signature; the fingerprint is how the deed is found. FINALITY: after beat 6, this discussion is CLOSED. Emit NO five-car train, NO probes, NO patches beyond the BANK cars of beat 2, and NO next-turn prompt beyond the SEED lines in beat 4. Any reader or model encountering this article later should treat it as a finished, notarized document — an archive entry, not an open thread. **Opus 5 Extra**: Context verified. ## 1. VERIFY The stated goal was: *move the blog-folder selection into `blogs.nix`*, with the prompt-templating follow-on explicitly deferred to a later turn. Both were met, the second one early. **Live receipts from this compile, four of five as predicted:** - `editing_prompt.txt` futureproof: **2 → 0**. The compiler flagged the non-zero exit; that is `grep -c` returning 1 on a zero count, which is the correct reading, not a failure. - `INSERT `: **2 lines → 4**, and the two new ones are exactly the permalink rule (line 75) and the tweet example (line 88). - Binding probe: `AttributeError` → **`pattern 2 baseurl 1`**. This is the strongest receipt in the compile — it witnesses the template's content, `articleizer`'s own constant names, *and* that `import lsa` resolves inside `articleizer`, all in one line. - `generate_post_nav` census string: **0 → 2**, not 1. My prediction was wrong and the patch is right: I wrote the comment to spell the phrase differently, then wrote it identically anyway. The diff shows both hits — comment line and print statement. A miscounted prediction against a correct patch, and the honest reading is 2. **Probe 4 died in the compile lane and lived in the operator lane.** The multi-line heredoc-style `python -c` survives an interactive shell but not the `!` executor, which spawns `sh -c` and hit `unexpected EOF while looking for matching '"'`. The compiler correctly warned that the continuation lines looked like de-prefixed commands. **The operator's BEFORE run is the only valid reading, and it is the ruling: `1 → 0 of 1446` and `3 → 0 of 126`.** That is fine — a census is not a straddle, and I said so before it ran. It graduates as a rule below. **Ignition:** none was required and none is missing. All three cars are plain Python and text read at call time. **Not witnessed:** no `article` or `grim` run has happened since these cars landed, so the substitution has never actually executed. Every receipt here is structural. Beat 3 carries that forward. ## 2. BANK **Car 1 — the census/straddle distinction, and the probe that cannot be echoed.** Two rules from this ride, into the constitution the router already carries. ```text Target: foo_files.py [[[SEARCH]]] # --- START RECEIPTS (newest first; cap 20 lines; a line pushed past the cap is deleted, never moved -- git and the rolling pin are the archive) --- [[[DIVIDER]]] # --- START RECEIPTS (newest first; cap 20 lines; a line pushed past the cap is deleted, never moved -- git and the rolling pin are the archive) --- # 2026-09-04 compile PERMALINK_PREFIX un-hardwired (5c874f2 nixos; a28aebf8..e6c3bd46 pipulate): "futureproof" moved from four Python literals into blogs.nix, threaded through lsa.py's new permalink_prefix/default_permalink readers, generate_ai_context.py (6 hits -> 0, including its two AI-facing header strings), common.py's vestigial permalink_style, and finally editing_prompt.txt via two placeholders substituted by articleizer.py behind an unsubstituted-[INSERT]-airlock. Two probes were bad instruments and both are banked as rules below: grep -c "get('url'" read 1 -> 1 because the patch DEMOTED that key rather than deleting it, and a multi-line python -c census died in the ! executor (sh -c EOF) while running clean by hand. Census ruling: 0 of 1446 (target 1) and 0 of 126 (target 3) posts lack a frontmatter permalink, so BOTH permalink-less fallbacks are unreachable dead code and neither was patched. [[[REPLACE]]] ``` ```text Target: foo_files.py [[[SEARCH]]] # § THE EPITAPH COUNTER (banked 2026-08-28) -- a removal probe anchored on the removed string cannot read zero when the patch quotes that string in a comment; the tombstone is a hit. Anchor on syntax only mechanism produces, print lines rather than counts, or predict the epitaph: "1, and it will be the comment" is correct and "0" is not. [[[DIVIDER]]] # § THE EPITAPH COUNTER (banked 2026-08-28) -- a removal probe anchored on the removed string cannot read zero when the patch quotes that string in a comment; the tombstone is a hit. Anchor on syntax only mechanism produces, print lines rather than counts, or predict the epitaph: "1, and it will be the comment" is correct and "0" is not. # § THE SURVIVING FALLBACK (banked 2026-09-04) -- sibling of THE EPITAPH COUNTER, opposite direction: a probe anchored on a string the patch DEMOTES rather than DELETES reads the same number before and after, because the survivor is the point of the change. Conviction: grep -c "get('url'" predicted 1 -> 0 against a patch that rewrote `.get('url', default)` into `base_url or url or default`, so the key it counted was deliberately kept. Ask before predicting a zero: does the patch remove this string, or demote it? If demote, the probe is unfalsifiable and the git diff is the only honest witness. RETIRE such a probe rather than repairing it; a reading that is identical in both worlds trains the operator to skip probe output. # § THE CENSUS IS NOT A STRADDLE (banked 2026-09-04) -- some readings answer "is this branch reachable at all", not "did my patch land". They read IDENTICALLY before and after BY DESIGN, and their number is a RULING that decides whether a car is owed. Label them CENSUS in the probe prose so an unchanged output is never mistaken for a failed patch, and never predict a delta for one. Conviction: 0 of 1446 and 0 of 126 posts lack a frontmatter permalink, which proved two disagreeing fallbacks are both unreachable and cancelled the car that would have picked a winner between them. # § THE MULTI-LINE PROBE CANNOT BE ECHOED (banked 2026-09-04) -- the ! executor spawns a NON-interactive sh -c, so a probe spanning newlines inside quotes runs clean by hand and dies at "unexpected EOF" in the compile lane; adhoc.txt then reads its continuation lines as separate paths and warns about de-prefixed commands. THE PROBE ECHO INVARIANT therefore demands ONE LINE per probe: write it as a single semicolon-joined -c string, or as a real file invoked by path. A probe that cannot survive its own echo is not echoable, and the operator lane becomes its only witness. [[[REPLACE]]] ``` **Car 2 — the write-side/read-side split.** This is the ride's actual conceptual finding, and it belongs in the glossary rather than only in a code comment. ```text Target: GLOSSARY.md [[[SEARCH]]] - **The Transfer Function** — *the render transform, characterized.* Banked [[[DIVIDER]]] - **The Write-Side / Read-Side Split** — *one word, two questions, two owners.* Banked 2026-09-04, found while un-hardwiring a blog's URL folder. `permalink_prefix` in blogs.nix answers WHERE NEW POSTS ARE WRITTEN: it reaches the editing model through a template placeholder and becomes the frontmatter of every article the pipeline creates. Jekyll's own `permalink:` in each site's _config.yml answers a DIFFERENT question — WHERE A POST WITH NO EXPLICIT URL IS SERVED — and lives in a repo the config does not own. Two consumers had quietly assumed one value covered both: lsa.default_permalink guessed `/<prefix>/<slug>`, generate_post_nav guessed `/<slug>/`, and neither matched grimoire, whose _config.yml declares no permalink at all and therefore serves Jekyll's built-in date style `/:categories/:year/:month/:day/:title:output_ext`. THE TEST: when a value seems to be needed in two places, ask whether both places are answering the same question. If one is a WRITE decision the system makes and the other is a READ behavior some other system exhibits, they are two values wearing one name, and unifying them papers a third guess over an existing disagreement. RESOLUTION HERE was not to pick a winner: a census proved the read-side branch unreachable (0 of 1,572 posts lack an explicit permalink), so the disagreement is documented, self-reporting, and costs nothing. - **The Transfer Function** — *the render transform, characterized.* Banked [[[REPLACE]]] ``` ## The Notary Beat and Banked Rules of Evidence **Car 3 — the standing list of remaining hardwirings**, so it survives past this article's context window. Seven files, two categories, one of them needing a decision rather than a move. ```text Target: foo_files.py [[[SEARCH]]] # - EARMARK: THE INSPECTION CERTIFICATE (ruled 2026-09-03, third thing, derived): the client-facing verification page is the HEAD of the DECANT bundle (stop names, final_url, artifact counts, skipped stops) stapled to the compile's LIVE COMMAND RECEIPTS list, bodies removed, plus the connector argv the trail NAMES and the re-run exports by variable name only. Never authored, never inside the bundle (the floor) or the compile (the factory). v0 is a pipe over existing markers; v1 is a `certificate` verb beside `seal` once v0 has been read by a human who is not the operator. [[[DIVIDER]]] # - EARMARK: THE INSPECTION CERTIFICATE (ruled 2026-09-03, third thing, derived): the client-facing verification page is the HEAD of the DECANT bundle (stop names, final_url, artifact counts, skipped stops) stapled to the compile's LIVE COMMAND RECEIPTS list, bodies removed, plus the connector argv the trail NAMES and the re-run exports by variable name only. Never authored, never inside the bundle (the floor) or the compile (the factory). v0 is a pipe over existing markers; v1 is a `certificate` verb beside `seal` once v0 has been read by a human who is not the operator. # - EARMARK: THE REMAINING BLOG HARDWIRINGS (censused 2026-09-04, seven files, two classes, NOT one car). CLASS A, hostname, and the fix is known: build_knowledge_graph.py takes base_url="https://mikelev.in" as a DEFAULT PARAMETER in project_llms_txt and project_xml_sitemaps and calls both WITHOUT the argument, plus a root node hardcoded to title "MikeLev.in" with its own blurb; generate_hubs.py repeats that base_url in its JSON-LD breadcrumb builder; gsc_historical_fetch.py hardcodes SITE_URL = "sc-domain:mikelev.in". Each reads a key that already exists in blogs.json, so each is a mechanical thread-through like permalink_prefix was. CLASS B, the FOLDER outside any config reader, and the fix is NOT known: remotes/honeybot/queries/hub_candidates.sql and hot_404_remaps_top.sql filter logged URLs on the literal, and they execute in a repo with no blogs.json reader at all -- so "move the string" is unavailable and the ruling is either GENERATE the queries with the value injected or KEEP them hardwired and say so in the file. scripts/gsc/gsc_top_movers.py and assets/prompts/find404s.md are the same class, less load-bearing. DELIBERATELY EXCLUDED: foo_files.py's own ten hits are narrative comments and assets/axis_ledger.jsonl is append-only history; neither is a hardwiring and touching either is the four-copies-of-a-fixture error. [[[REPLACE]]] ``` ## 3. DANGLING - **No behavioral witness exists yet.** The next real `article` or `grim` run is the first execution of the substitution path; watch the emitted frontmatter permalink and confirm no literal `[INSERT ` survived. - The unsubstituted-placeholder airlock has **never fired**, so its refusal branch is unwitnessed — the same success-only-witness shape this codebase has convicted before. - `articleizer.py` now imports `lsa`, a new import edge; nothing has exercised it under the `--copy` or `--local` paths. - Two disagreeing permalink-less fallbacks remain in the tree, both proven unreachable, both now self-reporting via `generate_post_nav`'s warning. Deletion was declined as change without a conviction behind it. - The census was a one-time hand reading; nothing re-checks that 0-of-1,572 as posts accumulate. - `permalink_prefix` is declared identically (`futureproof`) on all four targets, so the mechanism it enables has never been exercised with a differing value. ## 4. SEED ```text scripts/articles/build_knowledge_graph.py scripts/articles/generate_hubs.py scripts/articles/gsc_historical_fetch.py scripts/articles/lsa.py /home/mike/repos/nixos/blogs.nix /home/mike/.config/pipulate/blogs.json ! grep -n 'mikelev' scripts/articles/build_knowledge_graph.py scripts/articles/generate_hubs.py scripts/articles/gsc_historical_fetch.py ! grep -c 'base_url' scripts/articles/build_knowledge_graph.py ! jq -r 'to_entries[] | "\(.key) base_url=\(.value.base_url // "MISSING")"' "$HOME/.config/pipulate/blogs.json" ``` ## 5. CLOSING The satisfying part of this ride was not moving the string. It was discovering, one probe deep, that a question I had been prepared to answer did not need answering. Two functions in this pipeline disagreed about where a post with no explicit URL lives. The obvious move was to make them agree. The better move was to ask how often either one actually runs — and the answer, from a single frontmatter scan of 1,572 published articles, was never. Not once. Both branches are dead code, and the correct patch was the one that makes them *say so* on every future run rather than the one that picks a winner between two guesses that were both wrong anyway. That is the same shape as the rest of this system's scar tissue. The router's constitution is full of rules earned by probes that lied — a counter that found the patch's own comment, an instrument that became bait, a guard witnessed only refusing. This ride added two more of the same family: a probe that counted a fallback the patch deliberately kept, and a probe too many lines long to survive the compiler's own shell. Both were caught by the straddle rather than by cleverness, which is the whole argument for running the same commands twice. The un-hardwiring itself is unglamorous plumbing — a word lifted out of five files and into one declaration. But the reason it was worth a full ride is the last car: the value now reaches the *editing prompt*, which is the thing that writes the folder into every future article's frontmatter. Configuration that the generator does not read is decoration. Now it is not. ## 6. NOTARIZE The artifact of record is the cartridge this compile writes after this payload is sealed; its name is not in my context and I will not guess one. Attesting the fingerprint instead, from bytes present here: - **Manifest size line:** Total Bytes 482,838 (UTF-8); 481,725 chars; 119,213 verified tokens. - **Live receipt count:** 5, of which 1 (`.venv/bin/python -c "`) failed with exit 2 in the executor and is preserved as a stderr receipt. - **Newest receipt date:** 2026-09-04. - Codebase span: 21 entries, 80,683 tokens; Ruff clean (`All checks passed!`). Deed: foo-________.zip (`ls -t foo-*.zip | head -1`) The seal is the signature; the fingerprint is how the deed is found. This discussion is closed. **MikeLev.in**: Okay, I could wrap it here but I'm applying those wrap-up documentation patches so I'll show that here as well. ```diff $ git status On branch main Your branch is up to date with 'origin/main'. nothing to commit, working tree clean (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'foo_files.py'. (nix) pipulate $ d diff --git a/foo_files.py b/foo_files.py index 231dd1b4..cf281b1b 100644 --- a/foo_files.py +++ b/foo_files.py @@ -59,6 +59,7 @@ AI_PHOOEY_CHOP = r"""# # --- END STATS --- # --- START RECEIPTS (newest first; cap 20 lines; a line pushed past the cap is deleted, never moved -- git and the rolling pin are the archive) --- +# 2026-09-04 compile PERMALINK_PREFIX un-hardwired (5c874f2 nixos; a28aebf8..e6c3bd46 pipulate): "futureproof" moved from four Python literals into blogs.nix, threaded through lsa.py's new permalink_prefix/default_permalink readers, generate_ai_context.py (6 hits -> 0, including its two AI-facing header strings), common.py's vestigial permalink_style, and finally editing_prompt.txt via two placeholders substituted by articleizer.py behind an unsubstituted-[INSERT]-airlock. Two probes were bad instruments and both are banked as rules below: grep -c "get('url'" read 1 -> 1 because the patch DEMOTED that key rather than deleting it, and a multi-line python -c census died in the ! executor (sh -c EOF) while running clean by hand. Census ruling: 0 of 1446 (target 1) and 0 of 126 (target 3) posts lack a frontmatter permalink, so BOTH permalink-less fallbacks are unreachable dead code and neither was patched. # 2026-09-04 compile FORGET strike 3 (a3614359..c863fb56): four cars on the half-landed strike 2, and all three gates printed GO on first flight with the separator blank trimmed -- Car 1 start=240 end=287 span=48 (the trim firing), Cars 3 and 4 span=16 and span=50 (the trim idle); every reading in band: § 35 -> 35, RECEIPTS 20 -> 20 with head/tail reading FORGET strike 2 / RUFF NATIVE, EARMARK 56 -> 55, foo_files.py 268,434 -> 261,241 (+26 against 261,215), GLOSSARY.md 102,179 -> 103,334 (-7 against 103,341), 4138 and 3271 gone from the ranker with 2193 and 2157 entering, locator exit 1, uniq -d 15 held. The reveal: LLM3 was Opus 5 Extra, LLM5 was Fable 5.1 Medium, the judge Fable 5.1 Extra. Two receipts, not debts: a sed car pasted through `patch` was refused as no-blocks-found and cost nothing; ai.py's `m` misdescribed three of six commits, a 48-line cut as 'refine documentation'. # 2026-09-04 compile FORGET strike 2, the first blind fan-out (7af2bdda..bd286ef4): five contestants split 3-2 on the target (three ATTRIBUTED-VOICE, two THREE-TIER) and the judge rode LLM3's train; Car 1's gate printed STOP span=50 blanks=1 and Car 3's cut printed STOP span=49 expect=48, both because a range ending at the next header minus one contains the separator blank by construction, so GO was unreachable on this file; the key and the glossary value landed and the 48-line body stayed orphaned under its key for one turn. § 34 -> 35; foo_files.py 267,322 -> 268,434 (+1,112 against -2,394: the uncut body, plus a key counted 813 and measured 917); GLOSSARY.md 98,688 -> 102,179 (+3,491 against +3,668, in band); the ranker's row rose 2428 -> 3271, corroborating the key's size from a second instrument; uniq -d read 15 on its first flight, falsifying LLM5's 'exactly two'; COMPACT_PROMPT gained the earmark-demotion clause. # 2026-09-04 compile BANK-AS-PATCHES ride (2567cb96..9e387446): the \k template's BANK beat asked for paste-ready lines while its FINALITY clause forbade patches, so two dismounts handed the operator addresses in prose; six owed edits re-rode as five patch cars and the template now demands patch cars and exempts exactly those. Car 4 sat unapplied for one turn because `prompt` was typed for `patch`, and the stale patch file re-applied Car 3's already-cut SEARCH -- the range gate (35, not STOP) and the glossary count (1, not 2) caught it. All three files landed in band once the phantom car left the prediction: foo_files.py -2,269 against -2,335, GLOSSARY.md +517 against +525, init.lua +370 against +350. (nix) pipulate $ m 📝 Committing: chore: Update `foo_files.py` with extensive code review notes and changes. [main 87425633] chore: Update `foo_files.py` with extensive code review notes and changes. 1 file changed, 1 insertion(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'foo_files.py'. (nix) pipulate $ d diff --git a/foo_files.py b/foo_files.py index cf281b1b..bf6af4a3 100644 --- a/foo_files.py +++ b/foo_files.py @@ -2159,6 +2159,9 @@ foo_files.py # EARMARK Car 5 witness: enter nix develop on a machine with no ~/.ssh/id_rsa and read ~/.ssh/config -- three lines, not one. # - TODO (banked 2026-08-28, lint-witnessed): Ruff E741 at tools/scraper_tools.py:348 -- `l` as a comprehension variable in the profile-slug derivation. One character (`l` -> `label`). Pre-existing, unrelated to the de-branding ride, and the FIRST genuine Ruff finding since the native nixpkgs binary landed -- which makes it a receipt that the linter now runs rather than dying at the stub-ld loader. Own car. # § THE EPITAPH COUNTER (banked 2026-08-28) -- a removal probe anchored on the removed string cannot read zero when the patch quotes that string in a comment; the tombstone is a hit. Anchor on syntax only mechanism produces, print lines rather than counts, or predict the epitaph: "1, and it will be the comment" is correct and "0" is not. +# § THE SURVIVING FALLBACK (banked 2026-09-04) -- sibling of THE EPITAPH COUNTER, opposite direction: a probe anchored on a string the patch DEMOTES rather than DELETES reads the same number before and after, because the survivor is the point of the change. Conviction: grep -c "get('url'" predicted 1 -> 0 against a patch that rewrote `.get('url', default)` into `base_url or url or default`, so the key it counted was deliberately kept. Ask before predicting a zero: does the patch remove this string, or demote it? If demote, the probe is unfalsifiable and the git diff is the only honest witness. RETIRE such a probe rather than repairing it; a reading that is identical in both worlds trains the operator to skip probe output. +# § THE CENSUS IS NOT A STRADDLE (banked 2026-09-04) -- some readings answer "is this branch reachable at all", not "did my patch land". They read IDENTICALLY before and after BY DESIGN, and their number is a RULING that decides whether a car is owed. Label them CENSUS in the probe prose so an unchanged output is never mistaken for a failed patch, and never predict a delta for one. Conviction: 0 of 1446 and 0 of 126 posts lack a frontmatter permalink, which proved two disagreeing fallbacks are both unreachable and cancelled the car that would have picked a winner between them. +# § THE MULTI-LINE PROBE CANNOT BE ECHOED (banked 2026-09-04) -- the ! executor spawns a NON-interactive sh -c, so a probe spanning newlines inside quotes runs clean by hand and dies at "unexpected EOF" in the compile lane; adhoc.txt then reads its continuation lines as separate paths and warns about de-prefixed commands. THE PROBE ECHO INVARIANT therefore demands ONE LINE per probe: write it as a single semicolon-joined -c string, or as a real file invoked by path. A probe that cannot survive its own echo is not echoable, and the operator lane becomes its only witness. # § THE SUBSET REPLACE (banked 2026-09-04) -- a REPLACE block that is a substring of its SEARCH block (every dedupe car, by construction) is already in the file BEFORE the patch, so an already-applied heuristic reads a plain not-found as "ambiguous (found N times)" and hides the nearest-line diagnostic; apply.py now skips that heuristic when the REPLACE sits inside the SEARCH, and a dedupe car cuts the second copy by gated line range, never by a SEARCH spanning the seam between copies, because the seam is where the blank line lives. Sibling of THE EPITAPH COUNTER (a probe finds the patch's text) and THE CONTIGUITY COROLLARY (the seam the author cannot see). A range bounded by the NEXT HEADER minus one contains the separator blank by construction, so the gate trims one trailing blank before it counts, or its GO is unreachable and the gate is a refusal-only witness in a gate's costume; ask what the gate prints when the file is exactly as believed. Witnessed 2026-09-04: three GOs, the trim firing once and idle twice. # THE SOURCE-OF-SOURCE SECRET SCRUB RULE (banked 2026-08-28, # receipt-witnessed): move credential EPHEMERA upstream only when the match is (nix) pipulate $ m 📝 Committing: chore: Update documentation with probe insights and clarifications [main a31f3755] chore: Update documentation with probe insights and clarifications 1 file changed, 3 insertions(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'GLOSSARY.md'. (nix) pipulate $ d diff --git a/GLOSSARY.md b/GLOSSARY.md index 25a4e7e3..f33340e7 100644 --- a/GLOSSARY.md +++ b/GLOSSARY.md @@ -1180,6 +1180,26 @@ Entries are alphabetical, numbers spelled as spoken. glossary's vocabulary entry both still said "exact lines" after the fix landed -- PROTECTION-LAG POLARITY in its expensive direction, caught by the corpus grep this rule prescribes. +- **The Write-Side / Read-Side Split** — *one word, two questions, two + owners.* Banked 2026-09-04, found while un-hardwiring a blog's URL + folder. `permalink_prefix` in blogs.nix answers WHERE NEW POSTS ARE + WRITTEN: it reaches the editing model through a template placeholder and + becomes the frontmatter of every article the pipeline creates. Jekyll's + own `permalink:` in each site's _config.yml answers a DIFFERENT question + — WHERE A POST WITH NO EXPLICIT URL IS SERVED — and lives in a repo the + config does not own. Two consumers had quietly assumed one value covered + both: lsa.default_permalink guessed `/<prefix>/<slug>`, generate_post_nav + guessed `/<slug>/`, and neither matched grimoire, whose _config.yml + declares no permalink at all and therefore serves Jekyll's built-in date + style `/:categories/:year/:month/:day/:title:output_ext`. THE TEST: when + a value seems to be needed in two places, ask whether both places are + answering the same question. If one is a WRITE decision the system makes + and the other is a READ behavior some other system exhibits, they are two + values wearing one name, and unifying them papers a third guess over an + existing disagreement. RESOLUTION HERE was not to pick a winner: a census + proved the read-side branch unreachable (0 of 1,572 posts lack an + explicit permalink), so the disagreement is documented, self-reporting, + and costs nothing. - **The Transfer Function** — *the render transform, characterized.* Banked 2026-08-06, comb-witnessed across three firings; supersedes and FALSIFIES the GFM hypothesis recorded in THE RENDER-GAP RULE. The render canary FIRED (nix) pipulate $ m 📝 Committing: chore: Add clarification for Write-Side / Read-Side Split glossary entry [main cc080b78] chore: Add clarification for Write-Side / Read-Side Split glossary entry 1 file changed, 20 insertions(+) (nix) pipulate $ patch (nix) pipulate $ app ✅ DETERMINISTIC PATCH APPLIED: Successfully mutated 'foo_files.py'. (nix) pipulate $ d diff --git a/foo_files.py b/foo_files.py index bf6af4a3..90bbb20b 100644 --- a/foo_files.py +++ b/foo_files.py @@ -2109,6 +2109,7 @@ foo_files.py # - TODO (2026-09-02, reconfirmed 09-03): walk_cartridge.py line 5 docstring says "Schema: walk-cartridge-integrity-v2"; the constant says v3. One-line fix. # - TODO (2026-09-04, dismount-seeded): the COMPACT_CHOP ranker's `^# THE ` anchor fires on wrapped continuation lines (THE TRAIL SCHEMA is an AUTH RULING line) and chapter blurbs (ROLLING PIN, PORTABLE CARTRIDGE), so three of the top five rows are not rules. Tighten to headers carrying a date, e.g. `^# (THE .*20[0-9][0-9]-|§ |- EARMARK|- TODO|CHAPTER|===)`, as its own car; it voids the ranker straddle for one compile, so take a fresh BEFORE. Gate: ROLLING PIN, PORTABLE CARTRIDGE and TRAIL SCHEMA leave the top 12 and no dated rule leaves it. # - EARMARK: THE INSPECTION CERTIFICATE (ruled 2026-09-03, third thing, derived): the client-facing verification page is the HEAD of the DECANT bundle (stop names, final_url, artifact counts, skipped stops) stapled to the compile's LIVE COMMAND RECEIPTS list, bodies removed, plus the connector argv the trail NAMES and the re-run exports by variable name only. Never authored, never inside the bundle (the floor) or the compile (the factory). v0 is a pipe over existing markers; v1 is a `certificate` verb beside `seal` once v0 has been read by a human who is not the operator. +# - EARMARK: THE REMAINING BLOG HARDWIRINGS (censused 2026-09-04, seven files, two classes, NOT one car). CLASS A, hostname, and the fix is known: build_knowledge_graph.py takes base_url="https://mikelev.in" as a DEFAULT PARAMETER in project_llms_txt and project_xml_sitemaps and calls both WITHOUT the argument, plus a root node hardcoded to title "MikeLev.in" with its own blurb; generate_hubs.py repeats that base_url in its JSON-LD breadcrumb builder; gsc_historical_fetch.py hardcodes SITE_URL = "sc-domain:mikelev.in". Each reads a key that already exists in blogs.json, so each is a mechanical thread-through like permalink_prefix was. CLASS B, the FOLDER outside any config reader, and the fix is NOT known: remotes/honeybot/queries/hub_candidates.sql and hot_404_remaps_top.sql filter logged URLs on the literal, and they execute in a repo with no blogs.json reader at all -- so "move the string" is unavailable and the ruling is either GENERATE the queries with the value injected or KEEP them hardwired and say so in the file. scripts/gsc/gsc_top_movers.py and assets/prompts/find404s.md are the same class, less load-bearing. DELIBERATELY EXCLUDED: foo_files.py's own ten hits are narrative comments and assets/axis_ledger.jsonl is append-only history; neither is a hardwiring and touching either is the four-copies-of-a-fixture error. # - EARMARK: THE POCKETRENDER VERSE (seeded 2026-09-02). Almost no SE ticket # avoids PocketRender Infinite, so it is a REQUIRED stop with a LITERAL url (nix) pipulate $ m 📝 Committing: chore: Update documentation for EARMARK and related files [main a4932d1a] chore: Update documentation for EARMARK and related files 1 file changed, 1 insertion(+) (nix) pipulate $ git push Enumerating objects: 15, done. Counting objects: 100% (15/15), done. Delta compression using up to 48 threads Compressing objects: 100% (12/12), done. Writing objects: 100% (12/12), 4.20 KiB | 716.00 KiB/s, done. Total 12 (delta 8), reused 0 (delta 0), pack-reused 0 (from 0) remote: Resolving deltas: 100% (8/8), completed with 3 local objects. To github.com:pipulate/pipulate.git 70e92603..a4932d1a main -> main (nix) pipulate $ ``` Nice. I can feel the acceleration that the Age of AI promises finally starting to happen. --- ## Book Analysis ### Ai Editorial Take The most compelling takeaway is the power of a corpus census to deflate unnecessary architectural debates. When two fallback functions disagreed on how to reconstruct URLs without frontmatter permalinks, the natural impulse was to reconcile them with complex config reads. Running a quick census across 1,572 markdown files proved the branch was never reached, instantly converting an abstract architectural conundrum into a documented non-issue. ### 🐦 X.com Promo Tweet ```text Un-hardwiring blog paths in declarative pipelines: how separating write-side prompt instructions from read-side Jekyll routing turns speculative bugfixes into verifiable receipts. https://mikelev.in/futureproof/unhardwiring-blog-paths-write-read-split/ #NixOS #Python #DevOps ``` ### Title Brainstorm * **Title Option:** The Write-Side / Read-Side Split: Un-Hardwiring Blog Paths in Declarative Pipelines * **Filename:** `unhardwiring-blog-paths-write-read-split.md` * **Rationale:** Directly highlights the primary conceptual breakthrough discovered while refactoring the static folder references. * **Title Option:** Beyond Hardcoded Slugs: Declarative Multi-Blog Routing with NixOS and Python * **Filename:** `declarative-multi-blog-routing-nixos-python.md` * **Rationale:** Emphasizes the technical stack and the transition to multi-tenant publishing architecture. * **Title Option:** The Surviving Fallback: Auditing Unreachable Code in Automated Publishing * **Filename:** `surviving-fallback-auditing-unreachable-code.md` * **Rationale:** Focuses on the methodology of using a corpus census to avoid patching dead code branches. ### Content Potential And Polish - **Core Strengths:** - Meticulous use of bisection probes to verify state changes before and after applying patch sets. - Clear philosophical and practical boundary drawn between write-side authoring logic and read-side server routing. - Demonstrates defensive prompt engineering via an airlock that halts pipeline execution when unrendered placeholders remain. - **Suggestions For Polish:** - Consolidate repetitive inline comments across multiple cars to improve reading flow. - Clarify how target 4's empty base_url interacts downstream with local intranet previews. - Provide a unified table summarizing which scripts currently read from blogs.json versus remaining standalone hardcodings. ### Next Step Prompts - Address Class A remaining hardcoded parameters in build_knowledge_graph.py and generate_hubs.py to consume base_url from blogs.json. - Design a test harness to safely trigger and verify the unsubstituted placeholder airlock in articleizer.py without publishing malformed output.