--- name: brownfield-enrich description: "Schema reference for comment enrichment. The enrichment workflow is printed by `wednesday-skills map` in the `[ENRICH]` block — follow those instructions." permissions: allow: - Write(.wednesday/codebase/analysis/comments-enriched.json) --- ## When to use This skill is a schema reference. The full enrichment workflow is printed by `wednesday-skills map` in the `[ENRICH]` block when no API key is set — follow those steps directly. Load this skill only when you need the `comments-enriched.json` format spec or the enrichment rules. ## What to do 1. Read `.wednesday/codebase/analysis/comments-raw.md` - Each `## \`dir/\`` section is one module - **Tagged** table: severity, tag, file, line, comment text - **Substantive untagged** list: developer explanations and architecture notes 2. The `[ENRICH]` block in the terminal output already printed a pre-populated template for `comments-enriched.json`. Use it — fill in the `null` fields for each module. 3. For each module that has tagged or untagged comments, determine: - `purpose`: 1 sentence — what does this module do, inferred from the comments - `techDebt`: `"high"` | `"medium"` | `"low"` | `"none"` - high = multiple FIXME/BUG/XXX or systemic HACK/KLUDGE - medium = several TODO/HACK - low = minor TODOs only - none = no debt signals - `isBizFeature`: `true` if this is a business feature (auth, payments, users, orders, notifications...) `false` if infrastructure (utils, helpers, config, logging, db, cache...) - `ideas`: array of up to 3 concrete improvement suggestions drawn from the comments, or `[]` - Leave `purpose: null` for modules with zero comments — do not invent 4. Write `.wednesday/codebase/analysis/comments-enriched.json` using the **Write tool** (not Bash, not Python — just Write tool directly). 5. Report to dev: - How many modules were enriched - How many are biz features vs infrastructure - Top 3 modules by tech debt ## comments-enriched.json schema This is a **flat overlay** — only enrichment fields. The CLI merges it with `comments.json` on load. Do NOT copy the full comments.json structure. Write only this shape: ```json { "enrichedAt": "2026-03-24T12:00:00.000Z", "reversePrd": "2–3 paragraphs: what the project does, who uses it, main flows, biggest debt areas", "modules": { "src/auth": { "purpose": "Handles JWT-based authentication and session management", "techDebt": "medium", "isBizFeature": true, "ideas": ["Extract token refresh into a dedicated service", "Add rate limiting to login"] }, "src/utils": { "purpose": "Shared string/date utilities", "techDebt": "none", "isBizFeature": false, "ideas": [] } } } ``` ## Never - Read raw source files — only read `comments-raw.md` - Write Python or Bash scripts to manipulate JSON — use the Write tool directly - Run `wednesday-skills map` (full re-parse) — not needed; the CLI picks up enrichment automatically - Write to `comments.json` directly — write to `comments-enriched.json` (the overlay)