Connect any AI agent to Anchor.

Claude Web

1. Open Claude Web Connectors settings.
2. Click Add custom connector.
3. Name it Anchor and paste the server URL:
   https://mcp.anchor.cc/mcp
4. Click Add, then Connect. A Google sign-in window opens — use the account tied to your Anchor workspace.
5. In any new conversation, enable the Anchor connector from the tools panel and start prompting.

Try: "Create an Anchor table called Leads with columns for name, email, company, and status, fill in some data, then build a kanban app on top of it."

Claude Desktop

Chat Cowork Code

1. Open Claude Desktop and go to Customize → Connectors.
2. Click the + button, then choose Add custom connector.
3. Name it Anchor and paste:
   https://mcp.anchor.cc/mcp
4. Click Connect. A browser window opens for Google sign-in — approve to finish.
5. The Anchor connector is now available across all tabs — Chat, Cowork, and Code.

Claude Code

Terminal

Option A — one-line install

1. Run:
   claude mcp add --transport http anchor https://mcp.anchor.cc/mcp
2. Start Claude Code with claude and trigger any Anchor tool — your browser opens for Google sign-in on first use.
3. Verify the connection with /mcp inside Claude Code. Anchor should appear as connected.

Option B — edit the config file

1. Open ~/.claude/mcp.json (create it if missing).
2. Add Anchor under mcpServers:

{
  "mcpServers": {
    "anchor": {
      "type": "http",
      "url": "https://mcp.anchor.cc/mcp"
    }
  }
}

3. Restart Claude Code. The OAuth flow runs on first tool call.

ChatGPT

1. Open chatgpt.com and go to Settings → Connectors.
2. Choose Add connector → Custom MCP.
3. Fill in:
   • Name: Anchor
   • URL: https://mcp.anchor.cc/mcp
   • Authentication: OAuth (auto-discovered)
4. Click Save, then Connect and complete the Google sign-in.
5. In a chat, open the Tools menu, enable Anchor, and describe what you want built.

Codex

1. Open (or create) ~/.codex/config.toml.
2. Add an entry for Anchor:

[mcp_servers.anchor]
type = "http"
url  = "https://mcp.anchor.cc/mcp"

3. Run any Codex command. The first Anchor tool call triggers a browser OAuth flow — sign in with Google.
4. Confirm the connection with codex mcp list.

Gemini Enterprise

1. Sign in to the Gemini Enterprise admin console.
2. Go to Agents → Tools → MCP servers and choose Add server.
3. Fill in:
   • Name: Anchor
   • Transport: HTTP
   • URL: https://mcp.anchor.cc/mcp
   • Auth: OAuth 2.0 (auto-discovered)
4. Save. Attach Anchor to any agent template you want it available in — each end user authenticates with their own Google account the first time they call an Anchor tool.

Tip: Because each user signs in individually, Anchor permissions map cleanly to your existing team structure — no service accounts required.

Gemini CLI

1. Open ~/.gemini/settings.json (create it if missing).
2. Add Anchor under mcpServers:

{
  "mcpServers": {
    "anchor": {
      "httpUrl": "https://mcp.anchor.cc/mcp"
    }
  }
}

3. Launch Gemini CLI with gemini.
4. Type /mcp to list connected servers. Anchor should appear; the first tool call triggers a browser OAuth flow.

Grok

1. Open grok.com and go to Settings → Tools & connectors.
2. Choose Add MCP server.
3. Fill in:
   • Name: Anchor
   • URL: https://mcp.anchor.cc/mcp
4. Save and complete the Google sign-in when prompted.
5. Enable the Anchor tool from the chat toolbar before prompting.

Microsoft Copilot

1. Open Copilot Studio and pick (or create) the agent you want to extend.
2. Go to Tools → Add a tool → Model Context Protocol.
3. Register a new MCP server:
   • Name: Anchor
   • URL: https://mcp.anchor.cc/mcp
   • Authentication: OAuth 2.0 (dynamic registration)
4. Publish the agent. Users in your tenant sign in with their Google-linked Anchor account the first time they call a tool.

Note: MCP support is currently in public preview across the Copilot family. The connector path may differ slightly in Copilot for Sales, Service, and Finance SKUs.