Paper2Agent – transforming research papers into interactive AI agents

3 hours ago 1

Paper2Agent is a multi-agent AI system that automatically transforms research papers into interactive AI agents with minimal human input. Here are some Demos of the Paper2Agent-generated agent.

Automatically detects and runs all relevant tutorials from a research paper’s codebase.

⚠️ Prerequisites: Complete the installation & setup below before running Paper2Agent.

⏱️ Runtime & Cost: Processing time varies from 30 minutes to 3+ hours based on codebase complexity. Estimated cost: ~$15 for complex repositories like AlphaGenome using Claude Sonnet 4 (one-time cost).

cd Paper2Agent bash Paper2Agent.sh \ --project_dir <PROJECT_DIR> \ --github_url <GITHUB_URL>

Targeted Tutorial Processing

Process only specific tutorials by title or URL:

bash Paper2Agent.sh \ --project_dir <PROJECT_DIR> \ --github_url <GITHUB_URL> \ --tutorials <TUTORIALS_URL or TUTORIALS_TITLE>

For repositories requiring authentication:

bash Paper2Agent.sh \ --project_dir <PROJECT_DIR> \ --github_url <GITHUB_URL> \ --api <API_KEY>

Required:

--project_dir <directory>: Name of the project directory to create
- Example: TISSUE_Agent
--github_url <url>: GitHub repository URL to analyze
- Example: https://github.com/sunericd/TISSUE

Optional:

--tutorials <filter>: Filter tutorials by title or URL
- Example: "Preprocessing and clustering" or tutorial URL
--api <key>: API key for repositories requiring authentication
- Example: your_api_key_here

Create an AI agent from the TISSUE research paper codebase for uncertainty-calibrated single-cell spatial transcriptomics analysis:

bash Paper2Agent.sh \ --project_dir TISSUE_Agent \ --github_url https://github.com/sunericd/TISSUE

Scanpy Agent for Preprocessing and Clustering

Create an AI agent from the Scanpy research paper codebase for single-cell analysis preprocessing and clustering:

# Filter by tutorial title bash Paper2Agent.sh \ --project_dir Scanpy_Agent \ --github_url https://github.com/scverse/scanpy \ --tutorials "Preprocessing and clustering" # Filter by tutorial URL bash Paper2Agent.sh \ --project_dir Scanpy_Agent \ --github_url https://github.com/scverse/scanpy \ --tutorials "https://github.com/scverse/scanpy/blob/main/docs/tutorials/basics/clustering.ipynb"

Create an AI agent from the AlphaGenome research paper codebase for genomic data interpretation:

bash Paper2Agent.sh \ --project_dir AlphaGenome_Agent \ --github_url https://github.com/google-deepmind/alphagenome \ --api <ALPHAGENOME_API_KEY>

Python: Version 3.10 or higher
Claude Code: Install following instructions at anthropic.com/claude-code

Clone the Paper2Agent Repository

git clone https://github.com/jmiao24/Paper2Agent.git cd Paper2Agent
Install Python Dependencies
Install and Configure Claude Code

npm install -g @anthropic-ai/claude-code claude

🤖 How to Create a Paper Agent?

To streamline usage, we recommend creating Paper Agents by connecting Paper MCP servers to an AI coding agent, such as Claude Code or the Google Gemini CLI (it's free with a Google account!). We are also actively developing our own base agent, which will be released soon.

After pipeline completion, Claude Code will automatically open with your new MCP server loaded.

Manual Launch with Local MCP Server

To restart your agent later:

cd <working_dir> fastmcp install claude-code <project_dir>/src/<repo_name>_mcp.py \ --python <project_dir>/<repo_name>-env/bin/python

Manual Launch with Remote MCP Server Hosted on Hugging Face

To create a paper agent in Claude Code with the Paper MCP server of interest, use the following script with your own working directory, MCP name, and server URL:

bash launch_remote_mcp.sh \ --working_dir <working_dir> \ --mcp_name <mcp_name> \ --mcp_url <remote_mcp_url>

For example, to create an AlphaGenome Agent, run:

bash launch_remote_mcp.sh \ --working_dir analysis_dir \ --mcp_name alphagenome \ --mcp_url https://Paper2Agent-alphagenome-mcp.hf.space

✅ You will now have an AlphaGenome Agent ready for genomics data interpretation. You can input the query like:

Analyze heart gene expression data with AlphaGenome MCP to identify the causal gene for the variant chr11:116837649:T>G, associated with Hypoalphalipoproteinemia.

To reuse the AlphaGenome agent, run

Verify your agent is loaded:

or use \mcp inside Claude Code. You should see your repository-specific MCP server listed.

After completion, your project will contain:

<project_dir>/ ├── src/ │ ├── <repo_name>_mcp.py # Generated MCP server │ └── tools/ │ └── <tutorial_file_name>.py # Extracted tools from each tutorial ├── <repo_name>-env/ # Isolated Python environment ├── repo/ │ └── <repo_name>/ # Cloned repository with original code ├── claude_outputs/ │ ├── step1_output.json # Tutorial scanner results │ ├── step2_output.json # Tutorial executor results │ ├── step3_output.json # Tool extraction results │ └── step4_output.json # MCP server creation results ├── reports/ │ ├── tutorial-scanner.json # Tutorial discovery analysis │ ├── tutorial-scanner-include-in-tools.json # Tools inclusion decisions │ ├── executed_notebooks.json # Notebook execution summary │ └── environment-manager_results.md # Environment setup details ├── tests/ │ ├── code/<tutorial_file_name>/ # Test code for extracted tools │ ├── data/<tutorial_file_name>/ # Test data files │ ├── results/<tutorial_file_name>/ # Test execution results │ └── logs/ # Test execution logs ├── notebooks/ │ └── <tutorial_file_name>/ │ ├── <tutorial_file_name>_execution_final.ipynb # Executed tutorial │ └── images/ # Generated plots and visualizations └── tools/ # Additional utility scripts

Key Output Files and Directories

File/Directory Description

src/<repo_name>_mcp.py	Main MCP server file that Claude Code loads
src/tools/<tutorial_file_name>.py	Individual tool modules extracted from each tutorial
<repo_name>-env/	Isolated Python environment with all dependencies

Below, we showcase demos of AI agents created by Paper2Agent, illustrating how each agent applies the tools from its source paper to tackle scientific tasks.

🧬 AlphaGenome Agent for Genomic Data Interpretation

Example query:

Analyze heart gene expression data with AlphaGenome MCP to identify the causal gene for the variant chr11:116837649:T>G, associated with Hypoalphalipoproteinemia.

AlphaGenome_chatbot.mov

🗺️ TISSUE Agent for Uncertainty-Aware Spatial Transcriptomics Analysis

Example query:

Calculate the 95% prediction interval for the spatial gene expression prediction of gene Acta2 using TISSUE MCP. This is my data: Spatial count matrix: Spatial_count.txt Spatial locations: Locations.txt scRNA-seq count matrix: scRNA_count.txt

TISSUE_chatbot.mov

🧫 Scanpy Agent for Single-Cell Data Preprocessing

Example query:

Use Scanpy MCP to preprocess and cluster the single-cell dataset pbmc_all.h5ad.

🔗 Connectable Paper MCP Servers

AlphaGenome: https://Paper2Agent-alphagenome-mcp.hf.space
Scanpy: https://Paper2Agent-scanpy-mcp.hf.space
TISSUE: https://Paper2Agent-tissue-mcp.hf.space

@misc{miao2025paper2agent, title={Paper2Agent: Reimagining Research Papers As Interactive and Reliable AI Agents}, author={Jiacheng Miao and Joe R. Davis and Jonathan K. Pritchard and James Zou}, year={2025}, eprint={2509.06917}, archivePrefix={arXiv}, primaryClass={cs.AI}, url={https://arxiv.org/abs/2509.06917}, }

Read Entire Article