Two follow-ups from the code-tools v0.2.0 release (arch-audit):
- Calibration run. arch-audit was extracted from a single engagement (alpine-go-performance-starter, 2026-07-05/06). Its first use on a different repo —
astro-performance-starter is the natural candidate — should be treated as a test of the skill itself; revise the SKILL.md with whatever it gets wrong or misses.
constitution-bootstrap skill candidate (assessed worthwhile, not yet built): scaffolds the layered-AI-constitution pattern (docs/adr + CLAUDE.md layers + generated AGENTS.md spine + halt-on-violation quality gate) into an arbitrary repo. The pattern has now been hand-ported twice (astro → alpine-go). Proposes a draft for review rather than auto-generating.
🤖 Generated with Claude Code
Two follow-ups from the code-tools v0.2.0 release (arch-audit):
astro-performance-starteris the natural candidate — should be treated as a test of the skill itself; revise the SKILL.md with whatever it gets wrong or misses.constitution-bootstrapskill candidate (assessed worthwhile, not yet built): scaffolds the layered-AI-constitution pattern (docs/adr + CLAUDE.md layers + generated AGENTS.md spine + halt-on-violation quality gate) into an arbitrary repo. The pattern has now been hand-ported twice (astro → alpine-go). Proposes a draft for review rather than auto-generating.🤖 Generated with Claude Code