paper-revision — Claude Code skill: revise a paper from review feedback
Turns reviewer comments into concrete edits: each concern is tied to a section of the paper, then fixed. Handy once peer review comes back with revision requests.
npx skills add lingzhi227/agent-research-skills --skill paper-revision -a claude-codeWhat paper-revision does
Point it at the reviewer comments and your draft, either main.tex or the paper directory. The skill pulls out each individual concern, tags it as a major or minor revision, a question or a suggestion, and works out which sections it touches. Major concerns get handled first.
The output is a plan shaped like concern → section → action → new text or experiment. Actions range from a clearer rewrite to a missing experiment, an added ablation or statistical test, a restructure, or a citation. Edits stay local, so the surrounding structure survives. Any new number has to come from a run, not from the model's imagination, and revised text is flagged with \textcolor{blue}{...} so reviewers can find it.
Afterwards the skill re-checks the paper through self-review: are all concerns answered, did anything break, does the page count still satisfy the venue. A change summary follows, with every edit cross-referenced to the concern it resolves. Missing figures and tables get routed to the figure-generation and table-generation skills. Keep the pre-revision copy around, since the skill compares scores against it. Invented results are the one thing it cannot do—reviewers check.
What it helps with
- split a long reviewer report into individual concerns
- map each concern to the section it affects
- prioritize revisions so majors land first
- run a follow-up experiment a reviewer asked for
- produce a change summary tied to each concern
Example prompts for the agent
- Parse the review in
reviews.txt, revisemain.tex, and handle major concerns first - Plan the revisions from the review and mark changed text in blue
- Re-check with self-review, then write a summary of what changed
How to install paper-revision
1Skills CLI (Vercel)
npx skills add lingzhi227/agent-research-skills --skill paper-revision -a claude-code2Manually: put the files in place
~/.claude/skills/paper-revision/SKILL.mdCopy the whole skill folder; for one project use .claude/skills/paper-revision/
1Skills CLI (Vercel)
npx skills add lingzhi227/agent-research-skills --skill paper-revision -a codex2Manually: put the files in place
.agents/skills/paper-revision/SKILL.mdFor all projects use ~/.agents/skills/
1Skills CLI (Vercel)
npx skills add lingzhi227/agent-research-skills --skill paper-revision -a cursor2Manually: put the files in place
.cursor/skills/paper-revision/SKILL.mdCursor also reads .agents/skills/ and .claude/skills/
1GitHub CLI
gh skill install lingzhi227/agent-research-skills paper-revision2Manually: put the files in place
.github/skills/paper-revision/SKILL.md1Skills CLI (Vercel)
npx skills add lingzhi227/agent-research-skills --skill paper-revision -a gemini-cli2Platform CLI
gemini skills install https://github.com/lingzhi227/agent-research-skills --path skills/paper-revisionManually: put the files in place
.devin/skills/paper-revision/SKILL.mdThe legacy .windsurf/skills/ path still works
Skills CLI (Vercel)
npx skills add lingzhi227/agent-research-skills --skill paper-revision -a opencodeClaude.ai
Zip the folder and upload it in Settings → Capabilities → Skills
Step by step, with commands for each platform: How to install skills in Claude Code: five ways
Questions about paper-revision
How do I install paper-revision in Claude Code?
Run npx skills add lingzhi227/agent-research-skills --skill paper-revision -a claude-code. The skill file lands at ~/.claude/skills/paper-revision/SKILL.md.
What does paper-revision do?
It maps reviewer concerns onto sections of the paper, plans targeted edits, and applies them, running extra experiments where needed. Then it re-checks the paper and reports what changed.
What do I need before running it?
The reviewer comments and your current draft, as main.tex or a paper directory. Save a copy of the previous version first so you can compare scores before and after.
