Introduction
AI agents such as Claude Code, Codex, Claude and ChatGPT can work with BioLib on your behalf. Once connected, you can ask the agent to find an application for a task, explain how to run it, or check on your results.
Get started
The recommended way to connect your agent to BioLib is to install the BioLib plugin. The plugin connects your agent to the BioLib MCP server, and adds a skill that teaches the agent how to run applications with pybiolib. Install it from:
https://biolib.com/api/plugins.gitSee Claude Code and Codex for how to install it.
To connect Claude, ChatGPT or another agent that supports MCP directly, see MCP.
What the agent can do
The BioLib MCP server gives the agent read-only access to BioLib, as your user:
- Search applications: find applications by name or keyword.
- Get application details: read an application's description and arguments, with an example of how to run it.
- List results: list your most recent results, optionally filtered by state, with a link to each result.
The agent only sees the applications and results that your BioLib account has access to. To run applications, the agent can use pybiolib, which the BioLib plugin teaches it to do.
These capabilities are provided by the BioLib MCP (Model Context Protocol) server at
https://biolib.com/api/mcp/.
Sign in
The first time your agent connects to the MCP server, it asks you to sign in to BioLib:
- Your agent opens a BioLib page in the browser.
- Sign in to BioLib if you are not already signed in.
- Check the name of the application asking for access, and click Authorize. Only click Authorize if you initiated the request.
- You are sent back to your agent, which can now use BioLib.
Example prompts
Once connected, you can for instance ask your agent:
- "Find a BioLib app for predicting protein structure."
- "How do I run that app from Python?"
- "Did my latest BioLib results finish? Show me the ones that failed."
Still have a question?
If you have any questions that you can't find an answer to above, please reach out to the BioLib community.