Phase 0: docs, eval harness, agents and hooks

↗ view on GitHub · sloth-ninja · 2026-07-06 · 57833301

- CLAUDE.md: fill architecture/commands TODOs with verified module map,
  request flow, full env var registry, real dev/typecheck/lint commands
- docs/MIGRATION_SPEC.md: full US touchpoint inventory (file:line),
  CourtListener clean-removal plan, Companies House and legislation.gov.uk
  integration designs, local model (OpenAI-compatible) design, open questions
- evals/: golden-set runner (deterministic / citation hard gate / Opus-judged),
  live legislation.gov.uk citation resolution, judge wiring (claude-opus-4-8,
  structured JSON), seed cases + rubric; `npm run evals` / `npm run evals:smoke`
  from repo root (smoke verified green against the live API)
- .claude/agents/: test-writer, code-reviewer, eval-judge, terminology-auditor
- .claude/settings.json + hooks: PreToolUse guard (migrations/.env*/LICENSE),
  PostToolUse typecheck+lint+format on changed files, Stop hook running
  tests + evals:smoke

No feature code; no upstream source files modified except CLAUDE.md TODOs.

Co-Authored-By: Claude Fable 5 <noreply@anthropic.com>
Repository Sloth-ninja/JessicaOSS
Author sloth-ninja <ezanahaddis@gmail.com>
Authored
Parents 9f7c5b1b
Stats 25 files changed , +1861 , -14
Part of Add UK migration plan and evaluation scaffolding

Capture this commit into my fork

Download a Markdown prompt that tells Claude how to port this exact commit into your working tree. Run it via claude -p < capture-commit-57833301.md from inside the repo you want the change in.

⬇ Download capture-commit-57833301.md