🕵️♂️ Sherlock AI Plugin
The Ultimate Research & Implementation Assistant for Claude Code & Cursor
Sherlock is a curated collection of high-powered AI skills designed to transform your research workflow. It doesn't just read papers—it deduces, visualizes, and reconstructs them into working code.
From deep-diving into literature to converting methodology into Python, Sherlock equips your AI agent with the tools of a master detective and a master engineer.
🧰 The Essentials Kit (Features)
Sherlock equips your agent with these specialized skills:
🧩 1. Paper2Code (The Engineer)
Transforms academic theory into executable reality.
- 4-Stage Pipeline: Systematically converts research papers into code (Algorithm Extraction → Concept Analysis → Planning → Implementation).
- No Hallucinations: Forces a structured intermediate representation (YAML) before writing a single line of code.
- Reproducibility First: Prioritizes accuracy and paper fidelity over "clever" coding.
📚 2. Deep Research (The Detective)
The heavy lifter for comprehensive reports.
- Multi-Pass Drafting: Spawns parallel sub-agents to draft different sections of a report.
- Evidence Tracking: Maintains strictly cited evidence tables—no claim goes unsourced.
- High Fidelity: Produces professional-grade reports with strict formatting compliance.
🎨 3. Paper Comic (The Storyteller)
Explains the unexplainable through visual narratives.
- Visual Translation: Turns dense academic text into educational comics.
- Style Adaptive: Choose from Classic, Tech/Futuristic, Warm, or Chalkboard art styles.
- Gemini Powered: Uses the
genimg-gemini-webskill for consistent character consistency across panels.
🔬 4. Paper Analyzer (The Analyst)
X-Ray vision for PDFs.
- MinerU Integration: High-precision parsing of formulas, tables, and latex from PDFs.
- Style Rewrite: Can rewrite complex papers into "Storytelling", "Academic", or "Concise" formats.
- Metadata Extraction: Automatically pulls title, authors, and citations.
🖼️ 5. GenImg Gemini Web (The Artist)
The visual engine.
- Image Generation: Generates images via Google's Gemini Web.
- Multi-Modal: Handles text-to-image and image-to-text tasks.
- Session Awareness: Maintains context across multi-turn conversations for consistent output.
📐 6. Visual Architect (The Designer)
Blueprints for understanding.
- Schema Generation: Transforms methodology sections into structural visual schemas.
- Prompt Engineering: Generates high-precision prompts for DALL-E 3 or Midjourney based on paper logic.
- Layout Logic: Detects if a system is Linear, Cyclic, Hierarchical, or Parallel.
🚀 Quick Start
Installation
Option 1: CLI Install (Recommended)
Use npx skills to install skills directly:
# Install all skills
npx skills add proyecto26/sherlock-ai-plugin
# Install specific skills
npx skills add proyecto26/sherlock-ai-plugin --skill paper-comic genimg-gemini-web
# List available skills
npx skills add proyecto26/sherlock-ai-plugin --list
This automatically installs to your .claude/skills/ directory.
Option 2: Claude Code Plugin
Install via Claude Code's built-in plugin system:
# Add the marketplace
/plugin marketplace add proyecto26/sherlock-ai-plugin
# Install all marketing skills
/plugin install sherlock-ai-plugin
Option 3: Codex / Cursor / Copilot CLI (Agent Plugins)
This plugin follows the Agent Plugins 1.0 open standard, so the same repository installs on any conforming host:
| Host | Install |
|---|---|
| OpenAI Codex | codex plugin marketplace add proyecto26/sherlock-ai-plugin then codex plugin add sherlock-ai-plugin@sherlock-ai-plugin-marketplace (or pick it from /plugins) |
| GitHub Copilot CLI | copilot plugin marketplace add proyecto26/sherlock-ai-plugin then copilot plugin install sherlock-ai-plugin@sherlock-ai-plugin-marketplace |
| Cursor | Settings → Plugins → install from Git URL https://github.com/proyecto26/sherlock-ai-plugin (Agent Plugins format is detected from the root plugin.json) |
Which file each host reads:
plugin.json(repo root) — the portable Agent Plugins 1.0 manifest. Cursor and Copilot CLI read this directly and discoverskills/automatically..codex-plugin/plugin.json+.agents/plugins/marketplace.json— OpenAI Codex manifest and catalog..claude-plugin/plugin.json+.claude-plugin/marketplace.json— Claude Code manifest and catalog (Copilot CLI also reads this marketplace file).
All six skills are discovered from skills/ — no host manifest repeats the list,
so adding a skill is a one-directory change. Run bash scripts/check-manifests.sh
before releasing to confirm every manifest still agrees.
Copilot CLI note: installing from a working tree that contains
.git/can fail with access denied while Copilot copies the directory. Install from a clean checkout or from the marketplace.
Option 4: Clone and Copy
Clone the entire repo and copy the skills folder:
git clone https://github.com/proyecto26/sherlock-ai-plugin.git
cp -r sherlock-ai-plugin/skills/* .claude/skills/
Option 5: Git Submodule
Add as a submodule for easy updates:
git submodule add https://github.com/proyecto26/sherlock-ai-plugin.git .claude/sherlock-ai-plugin
Then reference skills from .claude/sherlock-ai-plugin/skills/.
Option 6: Fork and Customize
- Fork this repository
- Customize skills for your specific needs
- Clone your fork into your projects
Authentication
Some skills (like paper-analyzer) require API tokens or login sessions.
- MinerU: Export
MINERU_TOKENin your environment. - Gemini:
genimg-gemini-webneeds a signed-in Google session. Runnpx -y bun skills/genimg-gemini-web/scripts/main.ts --loginand complete sign-in in the Chrome window that opens.
Image generation requires being signed in. Google serves signed-out visitors a free Flash-tier model that declines to create images. Cached cookies expire, so if image generation stops working, re-run
--login.
Usage Examples
"Implement the code of this paper for me."
Triggers Paper2Code to read the PDF, plan the architecture, and write the Python implementation.
"Explain this complex transformer architecture as a comic."
Triggers Paper Comic to create a visual storyboard explaining the concept.
"I need a deep research report on the state of LLM reasoning."
Triggers Deep Research to crawl, analyze, and compile a multi-page cited report.
"Visualize the flow of data in this system."
Triggers Visual Architect to design a schematic diagram prompt.
📂 Structure
plugin.json # Agent Plugins 1.0.0 manifest (Cursor, Copilot CLI)
.codex-plugin/ # OpenAI Codex manifest
.claude-plugin/ # Claude Code manifest + marketplace
.agents/plugins/ # Codex / Copilot catalog entry
scripts/
└── check-manifests.sh # Asserts every host manifest agrees
skills/
├── deep-research/ # Report generation & evidence tracking
├── paper2code/ # Paper implementation pipeline
├── paper-comic/ # Educational comic generator
├── paper-analyzer/ # PDF parsing & style rewriting
├── genimg-gemini-web/ # Image generation backend
└── visual-architect/ # Visual schema design
🌟 Star History
💜 Sponsors
This project is free and open source. Sponsors help keep it maintained and growing.
Become a Sponsor | Sponsorship Program
🤝 Contribution
When contributing to this repository, please first discuss the change you wish to make via issue, email, or any other method with the owners of this repository before making a change.
Contributions are what make the open-source community such an amazing place to learn, inspire, and create. Any contributions you make are greatly appreciated ❤️.
You can learn more about how you can contribute to this project in the contribution guide.
👍 Credits
Happy vibe researching 💯
Made with ❤️ by Proyecto 26 - Changing the world with small contributions.
One hand can accomplish great things, but many can take you into space and beyond! 🌌
Together we do more, together we are more ❤️
No comments yet
Be the first to share your take.