seed-docslisted

# Seed Docs Let: σ := `.claude/stack.yml` config DOCS := `docs.path` from σ (default: `docs`) FMT := `docs.format` from σ (default: `md`) θ := stub threshold (< 30 non-blank, non-frontmatter lines OR ∃ `TODO:` markers) K := knowledge extracted from CLAUDE.md + codebase scan **Goal:** fill every stub doc so agents find actionable guidance, not placeholder text. **Idempotent** — skip files with ≥ 30 lines of real content (∄ TODOs). Safe to re-run. ``` /seed-docs → auto-discover DOCS from stack.yml /seed-docs --docs-path docs → explicit path /seed-docs --no-scan → skip codebase scan, use CLAUDE.md only ``` ## Phase 1 — Load Config Read σ (`cat .claude/stack.yml 2>/dev/null`). Record DOCS, FMT, `runtime`, `backend.framework`, `frontend.framework`, `backend.orm`. ¬σ → DOCS=`docs`, FMT=`md`. `--docs-path <p>` ∈ $ARGUMENTS → DOCS=p. `--no-scan` ∈ $ARGUMENTS → skip Phase 3. DOCS dir ∄ → → DP(A) **Run /init first** | **Create docs dir and seed** | **Cancel** - "Run /init" → explain Phase 7a creates stubs, exit. - "Create" → `mkdir -p {DOCS}/{architecture,standards,guides,processes}`, continue. ## Phase 2 — Read CLAUDE.md Read `CLAUDE.md` (∃) + any `@`-imported files it references. Extract into K: - **Purpose** — one-paragraph project summary - **Architecture style** — monolith | monorepo | microservices | plugin-based | other - **Module/folder structure** — src/ layout, package names, plugin dirs - **Naming conventions** — file, class, f