Flute Webhooks MCP is an MCP (model context protocol) server for the Flute platform.

It wraps the Flute Webhooks command-line tool. The server exposes API tools within several command groups. These cover endpoints, event types, delivery logs, and listeners. Endpoints can be listed, created, updated, deleted, or pinged. Deliveries can be listed, inspected, and retried.

A few commands are excluded on purpose, like interactive login. The headless listener is excluded, since it runs until canceled. Errors are reported clearly, with distinct kinds for each failure. This lets agents monitor webhook activity safely and predictably.

Credentials are never handled by the server itself. They are stored in the OS keychain or environment variables instead. Each server instance is pinned to one profile at startup. Profiles are either sandbox or production, never both at once. Credentials are never written to disk in plain text. They are also never exposed back to the calling agent.

For complete Flute Webhooks MCP details, including installation, operations, and help, see Flute Webhooks MCP.

An MCP (model context protocol) server exposes a set of API capabilities in a standard format.

It lets an AI agent drive the Flute command line tool. Otherwise, every AI app would need a custom integration for every tool.

Each tool call spawns a flute process and returns JSON. Errors come back in the CLI’s own error envelope too. Host apps like Claude Desktop or Codex can use this server directly. Agents never touch the Flute API directly this way.

These MCP capabilities are tools, resources, and prompts. Any MCP-compatible host or client can talk to it the same way. In the case of Flute Webhooks MCP, the server exposes the flute-webhooks CLI’s commands as MCP tools. Any MCP aware agent can call them this way. The agent never needs to know anything about the CLI itself.

MIT.