Journal Archive — Page 30 of 165

Repairing the FastHTML WebSocket Route in the Age of AI Learn how an upstream framework update broke chat socket routing in Python, and how a direct Starlette registration solved it.
The Machine That Has to Mean It: Auditing AI Narratives and Seams An important technical audit of AI onboarding narratives, examining how synthetic voices, unverified state claims, and silent chat-lane fallbacks create drift.
The Cartridge and the Codex: Zero-Dependency Replayability in the Age of AI Discover how portable, zero-dependency zip cartridges enable stateless AI context replay across disparate machines without complex deployment pipelines.
Falsifying the Vibe: Engineering Deterministic Codebase Upgrades in the Age of AI Learn how to transition from guesswork to falsifiable software engineering by applying deterministic bisection probes during complex framework upgrades.
Thermodynamics of Computation: From Maxwell's Demon to the Datacenter An important exploration of thermodynamic limits in computing, connecting Maxwell's Demon, Landauer's Principle, biological ratchets, and workflow ergonomics.
The Frontmatter Sandwich Ate the Industry: Bringing Integrity to the Agentic Web Discover how the industry converged on magic files like AGENTS.md, and how Pipulate turns static documentation into generated, unrottable code.
Stepping Up to FastHTML 0.14.9 Without Breaking a Sweat Learn how a conservative pip-compile roll-forward strategy and an AST-validated refactoring script successfully migrated 259 legacy components to FastHTML...
The Dayton MCP Wind Tunnel: Engineering the 3-Tier Evidence Ladder Discover how a rigorous 3-tier evidence ladder and a self-authored local fault harness turn protocol debugging into dependable engineering.
The Command Line Is Becoming the Context Line: Unix Verbs, ARexx Ports, and Local AI Workflows Discover how short Unix commands, append-only context files, and local-first automation create reproducible workflows for modern AI models.
Put the Whole Problem on the Bench: Engineering Deterministic AI Workflows Discover how to interlock local AI workflows, context cartridges, and deterministic validation to replace unguided agentic gambling with engineering rigor.