float

This MCP server exposes Float's complete API as a set of MCP tools, enabling AI assistants to manage Float data for projects, time tracking, resource allocation, and team coordination. The server is accessed via a Docker container and is designed to plug into MCP-enabled clients (such as Claude). You can use the four optimized tools to cover common workflows: manage-entity for CRUD operations across core entities (people, projects, tasks, departments, etc.), manage-project-workflow for project phases, milestones, and allocations, manage-time-tracking for time entries and PTO, and generate-report for analytics. Backwards compatibility tools are still available for legacy operations, preserving existing client configurations. To run, start the provided Docker image with your Float API key and connect your MCP client to the container’s exposed interface.

Prerequisites:

• Docker installed and running on your machine
• A valid Float API key with access to the Float API

1. Pull and run the Float MCP Docker image

# Replace YOURAPIKEY with your actual Float API key
export FLOAT_API_KEY=YOURAPIKEY

# Start the MCP server container (adjust as needed for your environment)
docker run --rm -i \
  -e FLOAT_API_KEY=$FLOAT_API_KEY \
  -e LOG_LEVEL=debug \
  ghcr.io/asachs01/float-mcp:latest

2. Alternative: customize environment variables You can override base URL or logging level via environment variables:

export FLOAT_API_BASE_URL=https://api.float.com/v3
export LOG_LEVEL=info

Then run the image as above (the env vars will be picked up by the container).

3. Verify connectivity

• Ensure the container starts without errors and the MCP client can load the float-mcp server configuration.
• In your MCP client, point the mcpServers.float-mcp configuration to the running container (see the example in the repository documentation).

4. Optional: run with a custom tag or registry If you pull from a different registry or tag, adjust the image name accordingly.

Tips:

• Ensure your FLOAT_API_KEY is kept secret; do not commit it to code repositories.
• The container exposes the MCP endpoint; you may need to configure port mappings in your environment if your MCP client expects a specific host:port.
• If you encounter rate-limiting, consider adjusting the LOG_LEVEL and leveraging the built-in exponential backoff in the server.
• For local development, you can use LOG_LEVEL=debug to get verbose logs for troubleshooting.
• The official Float MCP server is now recommended by Float; this community server is archived, so rely on the official solution for production usage when possible.

Environment variables:

• FLOAT_API_KEY: Your Float API key (required by the server)
• FLOAT_API_BASE_URL: Base URL for the Float API (default is https://api.float.com/v3)
• LOG_LEVEL: Logging level (error, warn, info, debug)

Common issues:

• Invalid API key or insufficient permissions: verify the key in the Float dashboard and ensure it has API access.
• Docker image pull failures: confirm network access and image name/tag correctness.
• MCP client configuration mismatch: ensure the client’s mcpServers.float-mcp points to the running container and uses compatible command/args.