Bonito MCP Server

A Model Context Protocol (MCP) server that wraps the Bonito AI Platform REST API. Use it from Claude Desktop, Cursor, or any MCP-compatible client to manage your multi-provider AI infrastructure.

Features

• Provider Management – Connect and verify AWS Bedrock, Azure OpenAI, GCP Vertex AI, OpenAI, Anthropic, and Groq
• Model Management – List, sync, and activate models across providers
• Unified Gateway – Send chat completions through Bonito's OpenAI-compatible gateway
• Projects – List and create projects (containers for agents, KBs, budgets)
• Agent Management – Create and execute BonBon and Bonobot agents
• Knowledge Bases – Create KBs, upload inline documents, and link KBs to agents for RAG
• Cost & Observability – Track costs per provider and view gateway logs
• Data Residency – Keep model calls in one geography (CA, US, EU) and see which providers can serve (get_data_residency, set_data_residency)
• Cloud Agent Inventory – List agents running in your own AWS and Google accounts (AgentCore, Bedrock Agents, Agent Registry, Agent Engine) with findings, and let Bonito agents call them through Bonito (list_cloud_agents, connect_cloud_agent, Pro+)

Total: 29 tools (22 after the 2026-06-07 audit pass; residency tools added 2026-09-25; list_cloud_agents and connect_cloud_agent added 2026-10-09; agent_relay 2026-10-10; update_gateway_key 2026-10-10). 1.4.1 fixes every tool failing with "ctx: Field required": the ctx parameter was untyped, so FastMCP exposed it as a string argument instead of injecting the request context; affected every release before it.

Audit notes (2026-06-07)

Tools added in the latest audit pass: list_projects, create_project, upload_to_kb, link_kb_to_agent. Without these, KBs were inert (no way to load docs via MCP) and agents couldn't be scoped to projects (which the post-Phase 1 create_agent flow requires).

Still missing from this MCP surface relative to the live platform — flag for the next pass:

• Access-token CRUD (bp- and bj- create/list/revoke)
• Image generation (POST /v1/images/generations)
• Video generation (POST /v1/videos + status + content)
• Agent-to-agent connections (connect_agents — handoff / escalation / data_feed / trigger)
• Agent updates (update_agent)
• Tier-aware model listing (current list_models is unfiltered)
• Origami tool surface (16 tools) is intentionally separate; do NOT mirror into MCP without product review.

Quick Start

Install from source

git clone https://github.com/ShabariRepo/bonito.git
cd bonito/mcp-server
pip install -e .

Set your API key

export BONITO_API_KEY="your-bonito-api-key"

Run (stdio – for Claude Desktop)

bonito-mcp

Run (SSE – for network clients)

bonito-mcp --transport sse --port 8080

Claude Desktop Configuration

Add to your claude_desktop_config.json:

{
  "mcpServers": {
    "bonito": {
      "command": "bonito-mcp",
      "env": {
        "BONITO_API_KEY": "your-bonito-api-key"
      }
    }
  }
}

Docker

docker build -t bonito-mcp .
docker run -e BONITO_API_KEY="your-key" -p 8080:8080 bonito-mcp

Available Tools

Environment Variables

License

MIT