research-papers

research-papers

A plugin for managing annotated research paper collections. Works with Claude Code, Codex CLI, and Gemini CLI.

This repo supports two distribution models:

• Claude Code marketplace repo via .claude-plugin/marketplace.json
• Script-based skill installer for Codex CLI and Gemini CLI via scripts/install_skills.py

What it does

This plugin provides skills for retrieving, reading, and annotating scientific papers into a structured, cross-referenced collection. Each paper gets:

• notes.md — Implementation-focused extraction (equations, parameters, algorithms)
• description.md — Three-sentence summary
• abstract.md — Verbatim abstract + interpretation
• citations.md — Full reference list + key citations for follow-up
• papers/index.md — Auto-generated list of paper directories in the collection

Skills

Skill	Description
paper-retriever	Download a paper PDF from arxiv, DOI, ACL Anthology, or sci-hub
paper-reader	Read a paper and extract structured notes (handles small/medium/large papers)
paper-process	Combined retrieve + read in one step
lint-paper	Audit paper directories for completeness and format compliance
reconcile	Cross-reference a paper against the collection bidirectionally
tag-papers	Add tags to untagged papers using their existing notes
research	Web research on a topic, structured findings report
extract-claims	Extract/enrich propositional claims from a paper into claims.yaml
make-skill	Create new skills from prompt files

Scripts

Repository-level installer utilities live in scripts/. Paper-collection helper scripts live in plugins/research-papers/scripts/.

Script	Description
lint_skill_frontmatter.py	Parse every SKILL.md frontmatter block and fail on invalid YAML
generate-paper-index.py	Rebuild papers/index.md and papers/tagged/ symlinks
cross-reference-papers.py	Find cross-references between papers in the collection
migrate-format.py	Convert legacy Tags: lines → YAML frontmatter, bold refs → wikilinks
generate_claims.py	Parse notes.md and generate claims.yaml for a single paper
batch_generate_claims.py	Generate claims.yaml for all papers in a directory
bootstrap_concepts.py	Deduplicate and group concept names from claims files

Claims Extraction Pipeline

The claims pipeline extracts machine-readable propositional claims from paper notes. The typical workflow:

1. generate_claims.py — Parses notes.md (parameter tables, equations, key findings) and produces a draft claims.yaml marked with stage: draft. Fast, deterministic, no LLM needed.
2. extract-claims skill — LLM-powered enrichment of the draft claims (adds context, fixes types, fills gaps). Can also create claims from scratch if no draft exists.
3. batch_generate_claims.py — Runs step 1 across an entire papers directory. Use --skip-existing to avoid re-processing.
4. bootstrap_concepts.py — Collects concept names from all claims files, deduplicates similar names, and outputs a unified concepts list.

# Single paper: generate draft, then enrich with LLM
uv run plugins/research-papers/scripts/generate_claims.py papers/Author_2024_Title
$extract-claims papers/Author_2024_Title

# Batch: generate drafts for all papers missing claims.yaml
uv run plugins/research-papers/scripts/batch_generate_claims.py papers/ --skip-existing

# After batch: deduplicate concepts across all claims
uv run plugins/research-papers/scripts/bootstrap_concepts.py papers/ --output concepts.yaml

Installation

Claude Code marketplace

Use the repo itself as a Claude Code marketplace source. The marketplace manifest lives at .claude-plugin/marketplace.json and currently exposes the plugin ID research-papers@research-papers-marketplace.

git clone https://github.com/ctoth/research-papers-plugin
cd research-papers-plugin
claude plugin marketplace add . --scope user
claude plugin install research-papers@research-papers-marketplace --scope user

If your clone already exists somewhere else, replace . with that repo path.

Script-based installer for Codex CLI and Gemini CLI

Use the bundled installer when you want the skills installed into Codex and/or Gemini user skill directories:

uv run scripts/install_skills.py doctor
uv run scripts/install_skills.py install --platform codex --platform gemini

What the installer does:

• discovers every skill directory under plugins/*/skills/*
• installs Codex skills into ~/.agents/skills
• installs Gemini skills into ~/.gemini/skills
• prefers whole-directory symlinks and falls back to copying if symlinks are unavailable

Useful variants: