docs-drift-review
Skill315 starsChanged 10 months agoApache-2.0View on GitHub
Analyze repository documentation for implementation drift, stale examples, missing coverage, spelling/clarity issues, and reorganization opportunities; produce a prioritized report with exact file references and proposed fixes.
1 other project uses a version of this.
Adopted by
Most recently taken up by
- openshift-online/rosa-boundarytheir own version, first seen Nov 1, 2026
Other versions1 project uses a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 1
- openshift-online/rosa-boundary★ 1same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository8 other files it gives its agents.
- AGENTS.mdAGENTS.md
- commit-helperSkill
- deep-analyseSkill
- design-syncSkill
- doc-syncSkill
- sanity-checkSkill
- theme-auditSkill
- worktree-managerSkill
The file
Show its rules and text
We split this skill into the rules it gives an agent
15 rules. About 0 are common, 15 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈15
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 11
Use this skill when the user asks to:
- Distinctive: line 19
verify whether a feature or interface changed without corresponding docs updates
- Distinctive: line 43
Prefer to inspect, in this order:
- Distinctive: line 58
Create a compact inventory:
- Distinctive: line 77
Prefer tests and schemas over comments when determining current behavior.
- Distinctive: line 144
Prefer direct evidence from code, tests, schemas, generated help, or examples.
- Distinctive: line 145
If behavior is ambiguous, say so explicitly.
- Distinctive: line 146
Do not infer runtime behavior solely from type names or comments.
- Distinctive: line 147
If tests contradict prose docs, treat that as a drift signal and note the contradiction.
- Distinctive: line 148
If docs contradict code but code appears accidental or buggy, do not decide policy.
- Distinctive: line 187
If the user asked for edits, propose a safe file-by-file sequence.
- Distinctive: line 191
Look especially for these drift patterns:
- Distinctive: line 215
Do not overcorrect for house style unless a local convention is clear.
- Distinctive: line 230
If the user wants changes:
- Distinctive: line 235
keep diffs minimal and reviewable
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Docs Drift Review
Purpose
Use this skill when the user asks to:
- compare docs with the implementation
- find stale or incorrect docs
- review README, guides, API docs, design docs, comments, or onboarding docs
- suggest documentation improvements
- identify spelling, grammar, naming consistency, or reorganization issues
- prepare a documentation cleanup plan
- verify whether a feature or interface changed without corresponding docs updates
Outcomes
Produce a report that:
- Identifies implementation drift between docs and code.
- Distinguishes confirmed issues from plausible-but-unverified concerns.
- Suggests precise improvements to wording, examples, spelling, structure, and navigation.
- Prioritizes findings by user impact.
- Recommends the smallest sensible patch set.
Non-goals
Do not:
- invent behavior not supported by code or tests
- rewrite all docs when a targeted patch is enough
- flag style preferences as defects unless they reduce clarity or conflict with repo conventions
- claim drift based only on naming differences without checking surrounding context
- modify files unless the user explicitly asks for edits
Inputs to inspect
Prefer to inspect, in this order:
AGENTS.mdand nearby agent instructions- top-level
README* - docs directories such as
docs/,documentation/,guides/,examples/ - API specs and schemas such as OpenAPI, JSON Schema, protobuf, CRDs, GraphQL schema
- source code implementing the documented behavior
- tests covering the documented behavior
- config files, CLI help text, and sample manifests
- changelogs, migration guides, release notes
Core workflow
Step 1: Build a docs map
Create a compact inventory:
- major doc files and their purpose
- feature areas covered
- likely source-of-truth files for each area
- stale-looking sections such as versioned commands, flags, env vars, API paths, screenshots, copied outputs, or step-by-step flows
Step 2: Build an implementation map
Identify:
- entrypoints
- public interfaces
- commands, flags, config keys, env vars
- API routes, request/response shapes
- feature gates, defaults, constraints
- examples and fixtures
- relevant tests
Prefer tests and schemas over comments when determining current behavior.
Step 3: Compare docs against implementation
Check for:
- renamed or removed commands, flags, env vars, config fields, APIs
- changed defaults
- changed prerequisites or setup steps
- outdated examples or sample outputs
- undocumented new behavior
- docs claiming support that code no longer provides
- docs omitting constraints, edge cases, or failure modes that matter in practice
For each suspected drift item:
- find exact evidence in docs
- find exact evidence in code/tests/specs
- decide: confirmed drift, likely drift, or insufficient evidence
Step 4: Review doc quality
Inspect for:
- spelling and grammar errors
- inconsistent naming of products, features, commands, files, or concepts
- ambiguous wording
- duplicate content across files
- weak information architecture
- sections in the wrong place
- poor scannability
- examples without explanation
- explanation without runnable examples
- missing cross-links
- outdated references to file paths or repo layout
Step 5: Produce a prioritized report
Organize findings into:
- Critical correctness issues
- Important missing or stale content
- Clarity and usability improvements
- Copyediting issues
- Reorganization proposals
For each finding include:
- severity: critical / high / medium / low
- confidence: confirmed / likely / uncertain
- docs file(s)
- implementation file(s)
- concise explanation
- recommended fix
- optional proposed replacement text
Severity guide
Use:
- critical for setup-breaking, security-relevant, migration-breaking, or API-breaking documentation errors
- high for materially misleading docs that waste time or cause wrong usage
- medium for missing context, omissions, stale examples, or structural issues that impair success
- low for spelling, grammar, naming consistency, and polish
Evidence rules
- Prefer direct evidence from code, tests, schemas, generated help, or examples.
- If behavior is ambiguous, say so explicitly.
- Do not infer runtime behavior solely from type names or comments.
- If tests contradict prose docs, treat that as a drift signal and note the contradiction.
- If docs contradict code but code appears accidental or buggy, do not decide policy; report the mismatch neutrally.
Output format
Use this structure:
Summary
- brief statement of overall docs health
- count of findings by severity
- top 3 issues
Findings
For each finding:
- ID
- Severity
- Confidence
- Docs evidence
- Implementation evidence
- Why this matters
- Recommended fix
- Proposed text (only when helpful)
Coverage gaps
List feature areas implemented but not documented.
Reorganization suggestions
Suggest merges, splits, moves, or index pages.
Quick wins
List the smallest high-value doc fixes.
Optional patch plan
If the user asked for edits, propose a safe file-by-file sequence.
Review heuristics
Look especially for these drift patterns:
- README says one command, CLI help exposes another
- docs list flags not present in parser definitions
- docs omit required env vars or credentials now enforced by code
- examples use old API paths or field names
- docs describe defaults that changed in config code
- migration guides no longer match actual upgrade path
- docs refer to directories or files moved in the repo
- generated artifacts or CRDs changed but prose docs did not
- screenshots or output snippets no longer match current UX or logs
Spelling and style heuristics
Flag:
- repeated words
- obvious spelling mistakes
- inconsistent capitalization of product names
- inconsistent code font for commands, filenames, flags, env vars
- headings that do not match the section content
- paragraphs that should be converted into steps, tables, or bullet lists
- long pages that need a TOC or subsection split
Do not overcorrect for house style unless a local convention is clear.
Reorganization heuristics
Suggest reorganization when:
- the same concept is explained in 3 or more places
- onboarding and reference material are mixed together
- migration guidance is buried in release notes
- examples are detached from the feature they demonstrate
- the top-level README is doing too much
- there is no single source of truth for configuration or API reference
When asked to edit
If the user wants changes:
- edit the highest-severity correctness issues first
- then fix stale examples and missing prerequisites
- then improve naming consistency and structure
- keep diffs minimal and reviewable
- preserve established terminology unless the repo clearly moved on
Suggested prompts
Examples of requests this skill should handle:
- "Review the docs for implementation drift."
- "Compare README and docs/ against the current CLI and config."
- "Find stale examples and propose fixes."
- "Audit API docs versus the server implementation."
- "Suggest doc reorganization and copyediting improvements."
Completion criteria
The task is complete when:
- the main doc surfaces were checked against the current implementation
- findings are evidence-based and prioritized
- the report clearly separates confirmed issues from guesses
- suggested fixes are concrete enough to implement
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

