Generate documentation from source code — docstrings, READMEs, API references. Triggers on: generate docs, write documentation, document this, create README, API docs, doc-gen.
Scanned 9/2/2026
Install to Claude Code
npx -y skills add Smart-AI-Memory/attune-ai --skill doc-gen --agent claude-codeInstalls into .claude/skills of the current project.
Are you the author of Doc Gen?
Add the live security badge to your README — it updates automatically with every re-scan.
[](https://www.skillsdirectory.com/skills/smart-ai-memory-doc-gen)More formats (shields.io, HTML) on the badges page.
---
name: doc-gen
description: "Generate documentation from source code — docstrings, READMEs, API references. Triggers on: generate docs, write documentation, document this, create README, API docs, doc-gen."
---
# Doc Gen
**IMPORTANT: Start your response with a context preamble.**
Call `help_lookup(topic="doc-gen", mode="preamble")` and
display the returned `preamble` text as a blockquote. Then
tell the user they can say "tell me more" for a step-by-step
guide, or answer the scoping questions below to proceed.
If the MCP call fails, fall back to:
> **Doc Gen** — Generates documentation from your source code — docstrings, README sections, API references.
## Scoping
Before running, ask:
1. **Target**: "Which file or module needs documentation?"
2. **Type**: "What kind of docs?"
- Docstrings — Add or update Google-style docstrings
- README — Generate a README section for a module
- API reference — Generate full API documentation
- Overview — High-level module explanation
## MCP Tools
| Tool | What It Does |
| ---- | ------------ |
| `doc_gen` | Generate documentation for a module |
| `doc_audit` | Check for stale or missing docs |
| `doc_orchestrator` | Full documentation maintenance pipeline |
## Execution
### Shared command workspace (preferred)
Open adapter `doc-gen` with the validated target and documentation type. Run
the bound read-only `doc_audit` action first and publish `audit_result` with
the exact proposed artifact paths. Present the proposal widget or Markdown;
only an explicitly confirmed `apply_docs` action authorizes those paths.
After `doc_gen`, publish `generation_result` with the exact files reported
from disk. The adapter hashes them independently and rejects writes outside or
different from the approved set. Run the returned `doc-import-audit`/symbol
reality probe and publish its exact command and outcome as
`validation_result`. A partial write or failed reality probe must say “did not
complete” and retain changed-file hashes for rollback. Preserve these gates
and receipts in compact text when the shared tools are unavailable.
For docstring generation:
```
doc_gen(source_path="<target module>")
```
For a full documentation audit first:
```
doc_audit(path="<target>")
```
Then generate docs for gaps found:
```
doc_gen(source_path="<gap file>")
```
For a complete pipeline (audit + generate + review):
```
doc_orchestrator(path="<target>")
```
## Output Format
**Prefer the rich panel.** If the tool response includes `panel_html`,
pass it to `mcp__visualize__show_widget` — the universal report panel
(title, score, findings/category sections; from
`attune.workflows.report_panel`). It shows an explicit "did not
complete" state on failure, never a false "clean". Fall back to the
markdown below when the widget surface is unavailable.
```markdown
## Documentation Report
**Files:** X | **Functions Documented:** Y | **Gaps:** Z
### Generated Docstrings
| File | Function | Status |
|------|----------|--------|
### Gaps Remaining
| File | Missing |
|------|---------|
```
## Help
After presenting results, call:
```
help_lookup(topic="doc-gen", mode="workflow_help")
```
If templates are returned, offer: "I have tips about
documentation generation — want to see them?"
## Follow-Up
After presenting results, offer:
- "Want me to apply these docstrings to the files?"
- "Should I audit the rest of the project?"
- "Want a README section generated from this?"
Is this your skill, or is something wrong with this listing? Request removal or report an issue. Author removals are honored within 72 hours.
No comments yet. Be the first to comment!