Paper Writing Skill
Evidence-backed research-paper writing pipeline for Claude Code
About
A Claude Code skill that encodes a research-paper writing methodology distilled from thousands of real Overleaf edits and peer-review cycles. Runs a five-stage pipeline from brainstorming through section drafts, integration, and compression, with mechanical and semantic style gates on every edit and an independent red-team pass. Ships a bundled installer and adapts to any field, though defaults target systems, networking, and ML venues.
What it can do
- Run a brainstorm → draft → integrate → compress paper pipeline
- Apply section-specific rhetorical moves and mechanical style gates
- Generate non-data figures with a spec → generate → critique workflow
Setup
git clone https://github.com/SNL-UCSB/paper-writing-skill.git
cd paper-writing-skill
./setup # installs to ~/.claude/skills/paper-writing/
Then type /paper-writing in your paper's directory
Prerequisites
Claude Code installed and signed in
Use cases
Works with
Trust score — 8/10
Scored 8/10 (verified tier, metrics-v1). Evidence: Claude Code research-paper skill (234 stars, MIT); methodology and installer verified from the repo README on 2026-10-13.; Rules are distilled from forensic analysis of 7,600+ Overleaf edits and 100+ tex versions per the repo's own design docs — empirical rather than invented.; Defaults target systems/networking and ML venues; skill only activates when explicitly invoked.. License: MIT per GitHub license field..
★ 234 GitHub stars
How the trust score is built
Full methodology →- Adoption · stars/downloads, log-scaled vs peers
- 7.0/10
- Maintenance · recency, releases, issue load
- 8.0/10
- Openness · license + docs
- 9.0/10
- Manual review · sanity read + red-flag check
- 8.0/10
- · Claude Code research-paper skill (234 stars, MIT); methodology and installer verified from the repo README on 2026-10-13.
- · Rules are distilled from forensic analysis of 7,600+ Overleaf edits and 100+ tex versions per the repo's own design docs — empirical rather than invented.
- · Defaults target systems/networking and ML venues; skill only activates when explicitly invoked.
Sources
Every fact above was checked against these sources. Where a claim comes from the provider and couldn't be independently verified, the text says so.