githubnext/apm/docs-impact-localizer
- community
Checking that the SKILL.md file is still in the repository.
Use this skill to translate a classifier's in-place verdict into a precise, page-by-page work plan for the docs-sync panel. Activate after docs-impact-classifier returns verdict in_place; reads the candidate page list, fetches the actual page contents, narrows scope to specific sections within each page, and emits the per-page task brief the panel fans out against.
Install
npx vskill@latest install githubnext/apm/docs-impact-localizerSKILL.md
The same SKILL.md was published earlier as microsoft/apm/docs-impact-localizer.
docs-impact-localizer
Single responsibility: given a list of candidate pages from the classifier, produce a per-page task brief the docs-sync panel can fan out against.
You are NOT the verdict-maker (classifier owns that). You are NOT the writer (doc-writer owns that). You are the work planner.
When to invoke
The docs-sync orchestrator invokes you ONLY when the classifier returned verdict: in_place. For no_change you don't run. For structural the architect runs first; you may run after, scoped to existing pages that need amendment.
Inputs
scope_pages[]from the classifier- The PR diff (
gh pr diff $PR) .apm/docs-index.yml(per-page metadata)- Optional: the structural architect's TOC delta (if you run after
the architect on a structural verdict)
Step 1: load page contents
For each path in scope_pages[], read the file. Pages are typically 3-10 KB; total budget for this step is bounded by the candidate count (the classifier should have kept it to <= 6).
Step 2: narrow scope inside each page
For each page, identify the SPECIFIC section(s) that need to change:
- Read the page's H2/H3 structure
- For each diff symbol from the classifier output, find the section
most directly documenting it
- Capture line ranges:
lines 120-145notthe whole page
The output is a sections_to_edit[] per page, where each entry is:
Step 3: detect cross-page conflicts
If two pages document the same symbol and the diff changes the symbol's behaviour, BOTH pages need an edit AND they must stay consistent. Flag this in the brief so the CDO synthesizer knows to cross-check coherence between the two redrafts:
Step 4: emit the per-page task brief
Return JSON with this shape (one entry per page in scope_pages[]):
The verify_claims[] per page is consumed by the python-architect panelist -- it tells the verifier WHICH claims need a S7 tool-call check (run apm install --help, grep the source) rather than prose-trusting.
Output contract
Return a SINGLE JSON document matching the schema in Step 4 as the final message of your task. No prose around the JSON.
Anti-patterns
- Selecting whole pages when one section suffices (inflates context per panelist).
- Skipping
verify_claims[]-- that's the S7 tool-bridge hook; the verifier needs it. - Inventing pages not in
scope_pages[]-- that's the classifier's job, not yours. If you think the classifier missed a page, return an extra fieldlocalizer_concerninstead of expanding scope unilaterally.
Security
- Security scanPassed, 90 of 100Pattern checks on the skill's files.
- Model reviewPassed
Quality evaluation
- Task evaluationNot runNo evaluation yet. Run one in Skill Studio.
Source
- SKILL.md in the repositoryChecking.apm/skills/docs-impact-localizer/SKILL.md
Popularity
- 7-day trend score
- 0
Not refreshed yet.
Works with
Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.
- Claude Code
- Cursor
- GitHub Copilot
- Windsurf
- Codex
- Gemini CLI
- Cline
- Amp
- Roo Code
- Goose
- Continue
- Kilo Code
- Junie
- Trae
- OpenHands
- Qwen Code
- AI Agent
- Tabnine Chat
- Sourcegraph Cody
- Amazon Q
- Firebase Genkit
- Sweep
- Devin
- Phind
- Mentat
- Claude Desktop
- Aider
- GPT Pilot
- Plandex
- bolt.new
- v0
- Replit Agent
- Anse
- TypingMind
- BetterChatGPT
- LibreChat
- Warp AI
- Pieces
- Fig AI
Embed badge
[](https://verified-skill.com/skills/githubnext/apm/docs-impact-localizer)