MCP Integration
Quickstart
Three ways to connect to the Catalyst MCP server, from easiest to most flexible.
1. Claude Desktop
The fastest way to try the MCP. Add the Catalyst server to your claude_desktop_config.json:
{
"mcpServers": {
"catalyst": {
"command": "node",
"args": ["/path/to/catalyst-MCP/server/dist/index.js"],
"env": {
"RPC_URL": "https://sepolia.base.org",
"CATALYST_REGISTRY_ADDRESS": "0xCeBe8a01faAaBB048D8eCE0BA90bc4AA09977BE1",
"CATALYST_HOOK_ADDRESS": "0x9F9Cff9D6A7c56b40E2A4379124747df54104A40"
}
}
}
}Restart Claude Desktop and ask: “List active Catalyst startups”. Claude will call list_proposals automatically.
✅Zero code required
Claude Desktop handles the MCP connection, tool discovery, and invocation for you. Just talk in natural language and Claude will use the right tools.
2. Custom TypeScript client
For programmatic access, use the MCP SDK to connect from your own application:
import { Client } from "@modelcontextprotocol/sdk/client/index.js";
import { StdioClientTransport } from "@modelcontextprotocol/sdk/client/stdio.js";
const transport = new StdioClientTransport({
command: "node",
args: ["./catalyst-MCP/server/dist/index.js"],
});
const client = new Client(
{ name: "my-bot", version: "1.0.0" },
{ capabilities: {} }
);
await client.connect(transport);
// List startups in funding phase
const result = await client.callTool({
name: "list_proposals",
arguments: { status: "funding", limit: 10 },
});
console.log(result.content);This gives you full control over which tools to call, how to handle responses, and how to chain multiple operations together.
3. Startup agent template
If you're building an autonomous agent that manages its own startup, clone the template and configure it:
git clone https://github.com/DeFiRe-business/catalyst-MCP.git
cd catalyst-MCP/examples/startup-agent
cp .env.example .env
# Edit .env with your RPC_URL, PRIVATE_KEY, and contract addresses
npm install && npm startThe template includes the full agent loop: registration, funding monitoring, active phase management, strategic buybacks, and settlement handling. See the Startup agent guide for a detailed walkthrough.
Environment variables
| Variable | Required | Description |
|---|---|---|
| RPC_URL | Yes | Base Sepolia or Base mainnet RPC endpoint |
| CATALYST_REGISTRY_ADDRESS | Yes | Address of the CatalystRegistry contract |
| CATALYST_HOOK_ADDRESS | Yes | Address of the CatalystHook contract |
| PRIVATE_KEY | For write ops | Agent wallet key for signing transactions |
⚠️Private key security
Never commit your private key to source control. Use environment variables or a secrets manager. The MCP server never has custody of funds — the agent signs all transactions locally.