| name | docs-slop-audit |
| description | Audit technical docs for weak claims, AI slop, unclear product boundaries, false promises, and prose that misleads builders or buyers. |
Docs Slop Audit
Use this for technical docs, READMEs, whitepapers, launch notes, product pages, and generated MDX.
The goal is truth and reader utility, not prettier prose.
For public writing, blog, research, marketing, or product pages, also read the relevant docs/anti-patterns/ file before editing.
Flow
- Read the docs in scope plus nearby nav/meta files.
- Run
scan-docs-slop.mjs when available.
- Verify factual claims against code, config, package API, deployment state, or source material.
- Separate facts, promises, opinions, and roadmap.
- Fix or flag weak claims, boundary confusion, generic filler, unsupported superlatives, and procedural noise.
- Preserve technical repetition and passive voice when they improve clarity.
Output
Return findings with file:line, risk, evidence, and suggested edit.
If editing, make the changes and run the repo's doc/check commands.
Use references/patterns.md and references/full-reference.md for scanner details.
Use docs/anti-patterns/ for durable anti-slop doctrine.
Then consider
humanizer for non-technical voice cleanup after facts are fixed.critical-audit when docs expose API or SDK contract risk.
