Skip to content
Back to search

research-superpower

Research superpowers for Claude Code

Safety scan

A static text scan, not a guarantee. Select the badge to see what was scanned and every finding.

Checked today
Badge for your README
findsafeskills safety rating

Markdown

[![findsafeskills safety rating](/api/listings/kthorn--research-superpower/badge.svg)](/listings/kthorn--research-superpower)

HTML

<a href="/listings/kthorn--research-superpower"><img src="/api/listings/kthorn--research-superpower/badge.svg" alt="findsafeskills safety rating" /></a>

Install

i

Install instructions from the README

As written by the repo's author. findsafeskills has not verified these steps.

Option 1: Install from Marketplace (Recommended)

# In Claude Code, add the marketplace
/plugin marketplace add https://github.com/kthorn/research-superpower

# Install the plugin
/plugin install research-superpowers@research-superpowers-marketplace

# Start asking literature search questions!
# Example: "Find papers on BTK inhibitor selectivity with IC50 data"
# Example: "Review literature on CRISPR gene editing in cardiomyocytes"
# Example: "What's known about MmpL3 inhibitors for tuberculosis?"

Option 2: Manual Installation

# Clone to your config directory
git clone https://github.com/kthorn/research-superpower ~/.config/research-superpowers

# Start Claude Code and tell it your literature search question

**Note:** These skills help you **find and review papers**. For analyzing experimental data or designing experiments, you'll need other tools.

What was scanned

The files the safety scan read, as written by the repository's author and shown as plain text. Lines with a finding are marked.

README.md · README · 189 lines

View this file on GitHub (read at the repository's current version: the scan did not record which commit it read)

  1. # Research Superpowers: Literature Search & Review
  2. Give Claude Code superpowers for **systematic literature searching and review**. Search PubMed, screen papers, extract data, traverse citations, and synthesize findings.
  3. **Focus:** Finding and reviewing scientific papers, not lab work or data analysis.
  4. ## What You Get
  5. A complete toolkit for **systematic literature reviews**:
  6. - **Literature Search** - PubMed and Semantic Scholar integration
  7. - **Smart Paper Screening** - Abstract scoring + deep dive for relevant data extraction
  8. - **Screening Rubrics** - Build and test custom relevance criteria collaboratively
  9. - **ChEMBL Integration** - Check if medicinal chemistry papers have curated SAR data (~99k papers)
  10. - **Open Access Finding** - Unpaywall API to find free versions of paywalled papers
  11. - **Citation Traversal** - Intelligent backward and forward citation following
  12. - **Large-Scale Screening** - Parallel subagent processing for 50+ papers
  13. - **Research Orchestration** - End-to-end workflow from query to synthesized findings
  14. ## Quick Start
  15. ### Option 1: Install from Marketplace (Recommended)
  16. ```bash
  17. # In Claude Code, add the marketplace
  18. /plugin marketplace add https://github.com/kthorn/research-superpower
  19. # Install the plugin
  20. /plugin install research-superpowers@research-superpowers-marketplace
  21. # Start asking literature search questions!
  22. # Example: "Find papers on BTK inhibitor selectivity with IC50 data"
  23. # Example: "Review literature on CRISPR gene editing in cardiomyocytes"
  24. # Example: "What's known about MmpL3 inhibitors for tuberculosis?"
  25. ```
  26. ### Option 2: Manual Installation
  27. ```bash
  28. # Clone to your config directory
  29. git clone https://github.com/kthorn/research-superpower ~/.config/research-superpowers
  30. # Start Claude Code and tell it your literature search question
  31. ```
  32. **Note:** These skills help you **find and review papers**. For analyzing experimental data or designing experiments, you'll need other tools.
  33. ## How It Works (Literature Review Workflow)
  34. 1. **Parse your literature question** - Extract keywords, data types, constraints
  35. 2. **Build screening rubric** - Define what makes papers relevant (optional, for large searches)
  36. 3. **Search PubMed** - Find candidate papers
  37. 4. **Screen abstracts** - Score papers for relevance (0-10)
  38. 5. **Deep dive on promising papers** - Fetch full text, extract specific data and methods
  39. 6. **Traverse citations** - Follow relevant references and citing papers (via Semantic Scholar)
  40. 7. **Synthesize findings** - Organize results in SUMMARY.md with structured extraction
  41. 8. **Track everything** - Maintain research session folder with findings, PDFs, and deduplication
  42. ## Project Structure
  43. ```
  44. research-superpowers/
  45. ├── skills/
  46. │ ├── getting-started/ # Introduction and workflow overview
  47. │ └── research/
  48. │ ├── answering-research-questions/ # Main orchestration
  49. │ ├── searching-literature/ # PubMed search
  50. │ ├── evaluating-paper-relevance/ # Abstract screening + deep dive
  51. │ ├── checking-chembl/ # ChEMBL SAR data lookup
  52. │ ├── traversing-citations/ # Citation network traversal
  53. │ └── finding-open-access-papers/ # Unpaywall integration
  54. ├── scripts/
  55. │ └── find-skills # Search for relevant skills
  56. └── hooks/
  57. └── session-start.sh # Auto-load at Claude Code startup
  58. ```
  59. ## Research Session Output
  60. Each research query creates a folder:
  61. ```
  62. research-sessions/2025-10-11-btk-inhibitor-selectivity/
  63. ├── SUMMARY.md # Main findings organized by relevance
  64. ├── papers-reviewed.json # Deduplication tracking
  65. ├── papers/
  66. │ ├── 10.1234_example.pdf
  67. │ └── ...
  68. └── citations/
  69. └── citation-graph.json # Citation relationships
  70. ```
  71. ## Skills Library
  72. ### Literature Search & Review Skills
  73. - **answering-research-questions** - Main orchestration workflow (search → screen → extract → synthesize)
  74. - **building-screening-rubrics** - Collaborative rubric design with test-driven refinement
  75. - **searching-literature** - PubMed API integration with query optimization
  76. - **evaluating-paper-relevance** - Two-stage relevance filtering (abstract + deep dive)
  77. - **subagent-driven-review** - Parallel screening for large searches (50+ papers)
  78. - **checking-chembl** - ChEMBL database lookup for medicinal chemistry papers
  79. - **traversing-citations** - Smart citation following via Semantic Scholar
  80. - **finding-open-access-papers** - Unpaywall API for finding free full text
  81. - **cleaning-up-research-sessions** - Safe cleanup of intermediate files after research complete
  82. ## Philosophy
  83. These skills focus on **systematic literature searching and review**:
  84. - **Precision over breadth** - Find papers with specific data you need, not just topically related
  85. - **Test-driven screening** - Build and validate relevance rubrics before bulk processing
  86. - **Smart traversal** - Only follow relevant citations to avoid exponential explosion
  87. - **Track everything** - Deduplicate papers, cache abstracts for re-screening, maintain provenance
  88. - **Check in regularly** - Report progress every 10 papers, checkpoint every 50
  89. - **Reproducible** - Save rubrics, queries, and methodology with each research session
  90. **What this is NOT:** These skills don't analyze experimental data, design experiments, or perform statistical analysis. They help you **find and review published literature**.
  91. ## Requirements
  92. - Claude Code CLI
  93. - Internet connection (for PubMed and Semantic Scholar APIs)
  94. - Optional: API keys for higher rate limits (free tier works for most use cases)
  95. ## Reducing Command Prompts
  96. When Claude runs API calls (curl commands), you may see approval prompts. To eliminate these for research sessions:
  97. ### Option 1: Pre-configure permissions (Recommended)
  98. Copy the template permissions file to your research project:
  99. ```bash
  100. # In your research project directory (e.g., research-sessions/YYYY-MM-DD-query/)
  101. mkdir -p .claude
  102. cp ~/.config/research-superpowers/.claude/settings.local.json.template .claude/settings.local.json
  103. ```
  104. This pre-approves:
  105. - All PubMed/NCBI API calls
  106. - All Semantic Scholar API calls
  107. - All Unpaywall API calls
  108. - DOI resolution calls
  109. - Basic file operations
  110. ### Option 2: Per-command approval
  111. When prompted, choose option 2: "Yes, and don't ask again for similar commands in this directory"
  112. **Note:** If this doesn't work (keeps prompting), use Option 1 instead. Claude Code may add specific commands instead of patterns.
  113. ## API Information
  114. **PubMed/E-utilities:**
  115. - Free, no API key required
  116. - Rate limit: 3 requests/second without key, 10 req/sec with key
  117. **Semantic Scholar:**
  118. - Free tier: 100 requests per 5 minutes
  119. - API key (free): 1000 requests per 5 minutes
  120. - [Get API key](https://www.semanticscholar.org/product/api#api-key)
  121. **Unpaywall:**
  122. - Free: 100,000 requests per day
  123. - No API key required (just provide email)
  124. - [Learn more](https://unpaywall.org/products/api)
  125. **ChEMBL:**
  126. - Free, no API key required
  127. - ~99,000 curated medicinal chemistry papers
  128. - Structured SAR data (IC50, MIC, structures, assays)
  129. - [Learn more](https://www.ebi.ac.uk/chembl/)
  130. ## Contributing
  131. This is an experimental project! Contributions welcome:
  132. - Try it with real research questions
  133. - Report issues and edge cases
  134. - Suggest new skills
  135. - Improve existing workflows
  136. ## License
  137. MIT License - see LICENSE file
  138. ## Acknowledgments
  139. Inspired by [Superpowers](https://github.com/obra/superpowers) by Jesse Vincent

Ratings and comments

Sign in to rate this skill.

Sign in

Free to use. If it’s helped, pass it on.