Skip to main content

​
Overview

Bananaflow exposes an MCP (Model Context Protocol) server that lets AI assistants like Codex, Claude Code, Claude Desktop, and Cursor access your workspace and documentation. Once connected, an assistant can list your agents, fetch agent definitions, and search Bananaflow docs on your behalf.

​
Prerequisites

  • A Bananaflow API key. Generate one at /api-keys. See API Keys.
  • Your Bananaflow MCP endpoint: https://app.bananaflow.ai/api/v1/mcp/
The endpoint is also shown in Platform Settings → MCP Server inside the Bananaflow UI.

​
Claude Code

Register Bananaflow as an MCP server with the Claude Code CLI:
Replace YOUR_API_KEY with the key you generated. Verify the server is connected:

​
Codex

Open Codex’s config file (~/.codex/config.toml) and add a bananaflow MCP server:
Replace YOUR_API_KEY with the key you generated. If you prefer to keep the API key out of config.toml, store it in an environment variable instead:
Then set the API key before starting Codex:
Verify the server is registered:

​
Claude Desktop

Open Claude Desktop’s config file (claude_desktop_config.json) and add the bananaflow entry under mcpServers:
Restart Claude Desktop after saving. The Bananaflow tools should appear in the tool picker.

​
Cursor and other MCP clients

Any MCP client that supports Streamable HTTP transport can connect with the same URL and header. Paste the configuration above into your client’s MCP settings file and replace YOUR_API_KEY.

​
Example prompts

Once the MCP server is connected, you can drive Bananaflow from your coding agent in plain English. A few prompts to try: Explore your workspace
  • “List my agents in Bananaflow.”
  • “Show me the definition of the agent called Lead Qualifier.”
  • “Which credentials and tools are configured in my Bananaflow workspace?”
  • “List the recordings from my most recent agent.”
Edit an agent
  • “In my Lead Qualifier agent, add a new agent node after the greeting that asks the caller for their budget, then routes to the existing qualification node.”
  • “Add an end-call node to Support Bot that triggers when the user says they are done, with a polite goodbye prompt.”
  • “Rename the intro node in Lead Qualifier to greeting and update any edges that reference it.”
Learn the platform
  • “What node types does Bananaflow support, and what fields does a knowledge_base node take?”
Agent edits are saved as a new draft version – your published agent keeps serving calls until you explicitly publish the draft from the Bananaflow UI.
The API key controls which workspace the assistant sees. Treat it like any other credential – do not commit it to source control or paste it into shared chats.