| name | docs |
| description | General-purpose documentation writing. Takes a topic (feature, module, or theme), investigates the code, and creates or updates documents under docs/ following the repository's existing documentation conventions. Use for architecture overviews, usage guides, and reference pages. |
| user-invocable | true |
| argument-hint | <topic or target path> |
docs
Write documentation that matches the target repository, not a template.
No pipeline.yml — dialogue-centric skills do not run under belt.
Step 1 — Survey
- Read the existing
docs/ tree: language, tone, heading style, index
files, directory taxonomy.
- Investigate the code the topic covers with Grep/Read. For unfamiliar
areas spanning 10+ files, dispatch
belt-agent:explorer subagents in
parallel (focus: flow or patterns).
Step 2 — Placement decision
Determine the target path and document type (architecture overview /
usage guide / reference) from the existing taxonomy. If either is still
ambiguous after the survey, ask via the frontier interview
(authoring-principles §4, typically one round) — placement, type,
audience, depth.
Step 3 — Write
- Follow the repository's documentation language and style; when the
repo has no stated policy, write in English.
- Verify every code statement in the document (file paths, command
names, config keys, API names) against the actual source — never
write from memory.
- Update cross-links: link related existing documents, and add the new
document to the index (README or docs index) when one exists.
Step 4 — Verify
- Check that every path referenced by the touched documents exists.
- Show the user the diff summary: files created/updated + one-line
purpose each.
Red flags
- Never restate what a referenced document already says — link it.
- Never document APIs or commands without checking they exist in the
code.
- Never invent a new docs/ subdirectory when an existing one fits.