Menu
Phase 3 — Deduplicate confirmed issues by writing a dedup.json mapping file. No files are moved or modified.
Install with Codex or Claude Copy this prompt, paste it into Codex, Claude, or another assistant, and let it review the skill page and install it for you.
Based on SOC occupation classification
| name | deduplicate |
| description | Phase 3 — Deduplicate confirmed issues by writing a dedup.json mapping file. No files are moved or modified. |
| user-invocable | true |
| allowed-tools | Read, Write, Glob, Grep |
Run from the repo root. Requires investigated issues from Phase 2.
Scan for all directories that have both issue.json and investigation.json with "verdict": "confirmed" (or "uncertain").
For each confirmed issue, read issue.json (title, quoted_text, description, spotter) and investigation.json (location, category).
Cluster by:
For each cluster, pick a canonical representative (the one with the best evidence or most specific location).
Write a single file: .review/dedup.json
{
"clusters": [
{
"canonical": "chunks/intro-00-a3f2bc/undefined_symbols",
"members": [
"chunks/intro-00-a3f2bc/undefined_symbols",
"chunks/intro-01-b7c1de/undefined_symbols"
],
"reason": "Same issue: undefined kappa, overlapping chunks"
},
{
"canonical": "holistic/missing-robustness",
"members": ["holistic/missing-robustness"],
"reason": "Singleton"
}
],
"cross_references": {
"chunks/model-00-c4d5/methodology_errors": ["chunks/results-00-e6f7/overclaiming"]
}
}
Important: Paths in dedup.json are relative to .review/. The canonical path is used by downstream phases to identify which issues to process.
Do NOT modify or move any issue.json or investigation.json files. The dedup mapping is a separate overlay.
git add -A
git commit -m "review: deduplication — <N> canonical issues from <M> total confirmed"
Report: how many canonical issues remain, how many were merged, how many have cross-references.
Phase 6 — Generate the REVIEW.md report by walking the append-only review tree. Optionally create GitHub Issues.
Add a custom issue spotter for the paper review. Describe what to look for and this will create the spotter file.
Phase 1 — Run the chunks x issue_spotters cross-product to find candidate issues. Uses edsl for parallel LLM review.
Draft author responses for each confirmed issue. Writes response.json next to each issue (append-only).
One-shot holistic referee review of the full paper. Issues are written to .review/holistic/ as append-only issue.json files.
Phase 2 — Investigate candidate issues. Writes investigation.json next to each issue.json with evidence and a confirm/reject verdict.