Menu
Phase 6 — Generate the REVIEW.md report by walking the append-only review tree. Optionally create GitHub Issues.
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 | publish-review |
| description | Phase 6 — Generate the REVIEW.md report by walking the append-only review tree. Optionally create GitHub Issues. |
| argument-hint | ["--no-github-issues"] |
| user-invocable | true |
| allowed-tools | Bash, Read, Write, Edit, Glob, Grep |
Run from the repo root. Requires the completed review tree in .review/.
Look for the roboree scripts path in the system prompt (it will say "roboree scripts are at: "). Use that path:
python <r2-scripts-path>/build_report.py \
--review-dir .review/ \
--config .review/config.json \
--output REVIEW.md
Then generate the styled HTML report:
python <r2-scripts-path>/build_html_report.py \
--review-dir .review/ \
--config .review/config.json \
--output REVIEW.html
Both scripts walk the append-only tree:
dedup.json)issue.json + investigation.json + votes.json + response.json for eachThe HTML version is self-contained (no external dependencies) with collapsible issue cards, severity badges, agreement bars, and model vote details. It prints well too.
Check if $ARGUMENTS contains --no-github-issues. If not, create issues on the repo.
Read paper_repo from .review/config.json. For each confirmed canonical issue with severity != not_an_issue:
gh issue create --repo <paper_repo> \
--title "<severity>: <title>" \
--body "<issue description with evidence>" \
--label "<severity>,<spotter>"
All links should be pinned to the paper_commit SHA from config.
Important: Ask the user for confirmation before creating GitHub Issues.
git add -A
git commit -m "review: complete"
Report: link to REVIEW.md, number of GitHub Issues created (if any).
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.
Phase 3 — Deduplicate confirmed issues by writing a dedup.json mapping file. No files are moved or modified.
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.