Model Context Protocol

Connect your AI agent to Chuhching

Chuhching speaks MCP — the open standard for giving AI agents real tools. ChatGPT and Claude.ai connect with OAuth; desktop and IDE agents can use a scoped token. Once connected, they can search, draft, manage your pipeline, and help run the workspace under your approvals.

Get started free Open agent setup

Server URL: https://mcp.chuhching.com/mcp

What your agent can do

Discover

Search podcasts, creators, and curators, and surface live opportunities that match your profile.

Pitch

Draft personalized, on-brand pitches and send them — then keep follow-up sequences moving.

Inbox

Read, triage, and reply to conversations without leaving your agent's chat window.

Pipeline

Create deals, move stages, and keep your CRM current as each conversation progresses.

Schedule

Book meetings and manage availability the moment a lead says yes.

Analytics & tasks

Pull performance summaries, list and create tasks, and check your wiki — all by asking.

Make your agent an expert, not just capable

The MCP gives your agent the tools. The free Chuhching skills pack gives it the judgment — the right workflow, order, and guardrails for every job, so the same task comes out right every time.

Proven workflows

Twelve skills covering onboarding, CRM, outreach, automations, second-brain, PR, podcast guesting, community, social, scheduling, and game launches.

Built-in guardrails

Compliance-checked sends, drafts-only social, quote-before-spend on paid discovery, and confirm-first on anything outward.

Works everywhere

Claude Code, Cursor, Windsurf, and Cline. Download once, drop into your agent's skills folder, and go.

Download the skills pack Read the setup guide

Three steps to connect

Choose your connection

ChatGPT and Claude.ai use Chuhching OAuth. Desktop and IDE clients can use a scoped Personal Access Token from Settings → Integrations → MCP.

Add the server to your client

Pick your client below. Paste the server URL and sign in for OAuth, or paste the generated token into the desktop/IDE config.

Verify, then expand access

Start with a read-only workspace-list call. Add write or admin scopes only when a task needs them, and keep sends and publishing approval-gated.

Set up your client

Choose your client below. OAuth clients connect with the server URL and a Chuhching sign-in; desktop and IDE clients use the token placeholder shown in their config.

Connect your client

Server URL: https://mcp.chuhching.com/mcp

Each snippet has a <YOUR_TOKEN> placeholder — swap in a token from Settings → Integrations → MCP.

The desktop app launches MCP servers locally, so it reaches our hosted endpoint through the open-source mcp-remote bridge (npx fetches it on first run — no install). Newer builds with native HTTP transport can use the simpler second form.

Config file

macOS~/Library/Application Support/Claude/claude_desktop_config.json
Windows%APPDATA%\Claude\claude_desktop_config.json
Linux~/.config/Claude/claude_desktop_config.json

Open Claude → Settings → Developer → Edit Config (or open the file above).
Paste the snippet and replace <YOUR_TOKEN> with your token.
Fully quit and reopen Claude Desktop. PitchIQ's tools appear in the tool list within seconds.

mcp-remote bridge — works on every version (recommended)

{
  "mcpServers": {
    "pitchiq": {
      "command": "npx",
      "args": [
        "-y",
        "mcp-remote",
        "https://mcp.chuhching.com/mcp",
        "--header",
        "Authorization: Bearer <YOUR_TOKEN>"
      ]
    }
  }
}

Native HTTP — only if your build supports HTTP transports

{
  "mcpServers": {
    "pitchiq": {
      "type": "http",
      "url": "https://mcp.chuhching.com/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_TOKEN>"
      }
    }
  }
}

The bridge form needs Node.js installed (for npx). If tools don't appear, confirm Node is on your PATH and that you fully restarted the app.

Claude Desktop MCP docs ↗

About your token

Tokens are scoped — grant only the access an agent needs (read-only, or specific write scopes like outreach or pipeline). We store only a hash, never the plaintext, so we can't show a token twice; if a machine is compromised, revoke it and generate a new one. Treat tokens like passwords — never commit one to a public repo.

Give your agent a CRM

Start free, connect with OAuth or a scoped token, and let your AI agent help with discovery, drafts, and pipeline work without leaving the chat.

Get started free