Favro users have built and published open-source MCP (Model Context Protocol) servers that allow you to connect Favro to AI tools like Claude Code, Cursor, and others.
With an MCP server, you can interact with your Favro data through natural language. Create cards, update boards, pull status reports, and more, directly from your AI assistant.
Where to find them
Example 1:
Community made and maintained Favro MCP servers are available on GitHub:
Each repo includes setup instructions. Follow the README to get connected.
Example 2:
Pipedream, a Workday company, is hosting their own Favro MCP server here:
Important: use at your own risk
These servers are built and maintained by the community, not by Favro. We have not reviewed, tested, or vetted them for security, reliability, or accuracy.
By using a community-built MCP server, you accept that:
Favro is not responsible for any issues, data loss, or unintended changes caused by third-party servers.
You should review the source code and understand what permissions you're granting before connecting any tool to your Favro organization.
Support for these servers is handled by their respective maintainers, not by Favro's support team.
In short: these are community projects. We think it's great that they exist, but you use them at your own risk.
Our recommended approach for working with agents: build skills backed by the Favro API instead of using MCP servers
If you want the most control and reliability when connecting AI agents to Favro, we recommend building skills backed by scripts that use the Favro API directly, instead of using an MCP server. This gives you full control over what data is read and written, access to more of Favro's feature set, no dependency on third-party code you don't control, and typically better context management and token efficiency by fetching only the data needed for each task.
Example prompt:
"Create a skill backed by scripts that use the Favro API to triage bugs on a specific Favro board. If the user does not provide a collection or board, ask for them. Review new cards, categorize them, set priority, and return a summary of what was triaged. Use direct Favro API calls so only the necessary data is fetched for each task. Keep secrets in environment variables, and include a simple CLI and README. If there's anything unclear, ask me to clarify one item at a time"
This is especially relevant for production workflows, sensitive data, or anything where you need confidence in exactly what's happening under the hood.
Learn more here: Favro API documentation