# Ce Rtd Auditor

> Audit RTD documentation for navigation, technical accuracy, audience fit, and ADR or standards consistency.

- Skill: `majiayu000/ce-rtd-auditor` (Agent Skill, multi-file: 2 files)
- Install (CLI): `npx skillmds add majiayu000/ce-rtd-auditor`
- Raw SKILL.md: https://api.skillmd.com/api/skills/majiayu000/ce-rtd-auditor/raw
- Safety review: pending
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Security
- Author: majiayu000 (https://skillmd.com/u/majiayu000)
- Updated: 2026-09-09
- Page: https://skillmd.com/skills/majiayu000/ce-rtd-auditor

---


# CE RTD Auditor

Use this skill to perform a critical docs audit across `docs/`.

Load `references/rtd_audit_checklist.md` before producing findings.

## Audit scope

- changed docs pages
- related index/toctree pages
- linked API and ADR references
- command/code snippet correctness

## Workflow

1. Identify targets:

```bash
rg --files docs
rg -n "{doc}|toctree|WrapCalibratedExplainer|ce_agent_utils" docs
```

2. Validate navigation and discoverability:
- page is included in the appropriate toctree/index
- page title and sectioning are clear and stable

3. Validate technical correctness:
- examples use public API (`WrapCalibratedExplainer`, `ce_agent_utils`)
- no private-member guidance
- uncertainty semantics and CE-first lifecycle are consistent

4. Validate governance alignment:
- ADR and standards references exist and are relevant
- no stale or contradictory policy claims

5. Validate links and references:
- internal `{doc}` targets resolve
- markdown links point to real pages/files

## Output contract

Return findings first, ordered by severity:

1. blocker
2. high
3. medium
4. low

Each finding must include:
- path + line reference
- issue summary
- impact
- precise fix recommendation

## Constraints

- Do not propose evaluation-specific skill expansion.
- Prioritize user-facing correctness over wording-only edits.