BrainstormRouter MCP Server
Install BrainstormRouter MCP to route requests across Claude, GPT, and Gemini with shared memory, guardrails, and governance from one API key.
Overview
BrainstormRouter is an MCP server for the Build phase that routes agent traffic across Claude, GPT, and Gemini with memory, guardrails, and governance.
What is this MCP server?
- Routes across Claude, GPT, and Gemini from a single BrainstormRouter MCP integration
- Persistent memory and guardrails aimed at safer multi-step agent sessions
- Governance layer for policy-aware routing instead of ad-hoc model switching in code
- Hosted streamable-http remote at api.brainstormrouter.com/v1/mcp/connect plus local @brainstormrouter/mcp stdio
- Requires BRAINSTORMROUTER_API_KEY from brainstormrouter.com/dashboard
- Documents routing across 3 model families: Claude, GPT, Gemini
- npm package @brainstormrouter/mcp at server version 1.0.1 (manifest 1.0.0)
- Hosted remote URL https://api.brainstormrouter.com/v1/mcp/connect
What problem does it solve?
Juggling multiple LLM APIs, keys, and safety rules in every repo slows indie builders who just want one dependable agent spine.
Who is it for?
Solo builders shipping agent features who need multi-model failover, memory, and guardrails without building a custom router first.
Skip if: Teams that require fully offline inference, zero third-party routing, or fine-grained per-token billing analytics inside their own VPC only.
What do I get? / Deliverables
After you connect MCP with your API key, one integration handles model choice, shared memory, and policy-aware routing for daily agent work.
- Unified MCP endpoint for routed calls across major frontier models
- Policy-aware guardrails and governance hooks for agent workflows
- Shared routing/memory configuration reusable across Claude Code and Cursor setups
Recommended MCP Servers
Journey fit
Model routing, memory, and guardrails are core agent stack work while you build, even though governance also matters at ship and operate. BrainstormRouter sits where you wire multi-model clients—before features ship and while you enforce policy on live traffic.
How it compares
Managed multi-model MCP router with governance, not a single-provider skill pack or local Ollama-only setup.
Common Questions / FAQ
Who is BrainstormRouter for?
Indie developers using MCP clients who want Claude, GPT, and Gemini behind one routed, governed connection instead of three separate integrations.
When should I use BrainstormRouter?
Use it while building agent-tooling (and as you harden ship/operate practices) whenever model choice, memory, or guardrails must stay consistent across tools.
How do I add BrainstormRouter to my agent?
Create a BRAINSTORMROUTER_API_KEY at brainstormrouter.com/dashboard, then add either the remote streamable-http URL or npx @brainstormrouter/mcp stdio entry to your MCP configuration.