$ tail -f ~/blog | grep #ai
Posts filed under #ai.
2026-10-06 · 5 min read Handrail: write your guardrails once, enforce them everywhere Share guardrails across Claude Code, Codex, and oh-my-pi with a single rule definition, evaluated by native hooks at event time. # ai · opensource · productivity 2026-09-26 · 4 min read Three agentic skills I wish I'd started using sooner improve-tests, debrief-skill, and retro: three agent skills that handle the maintenance work that always loses to something urgent, from slow test suites to friction in skills and tooling. # ai · productivity · testing 2026-09-23 · 3 min read ArchSpec: when your AI agent doesn't know your rules, make the rules unbreakable ArchSpec turns Rails architecture conventions into rules that CI enforces, so code from AI agents, new hires, and seniors alike has to respect your layer boundaries. # ruby · rails · ai · architecture 2026-09-22 · 6 min read Agent shortcuts: aliases for your AI coding agent Shell-style shortcuts for Claude Code and Codex: commit, push, open pull requests, fix CI, and keep common agent workflows predictable. # ai · productivity · git 2026-09-21 · 7 min read System 1 AI models: the idea that existed before Jev, and why you should use it anyway System 1 models separate structured decisions from text generation. A look at Jev, earlier research, open models, and five possible applications. # ai · classification · llm 2026-08-06 · 14 min read Your agent writes Python. The Ruby rule cuts that by a third. A controlled comparison of Ruby, Python, and shell scripts written by a coding agent, measuring output size, correctness, and the effect of instructions. # ruby · ai · python · productivity 2026-08-04 · 19 min read I wrote down 61 rules, then failed my own repos How oss-kit uses 61 repository maintenance rules and nine agent skills to audit documentation, security, CI, releases, and community files. # opensource · ai · devops · showdev