Short answer

InstaChime MCP gives approved AI tools narrow, auditable access to lead-response data.

Connect a compatible Streamable HTTP client to https://instachime.com/api/mcp with a one-time workspace token. The client sees only the tools allowed by that token's scopes, and customer-facing outreach remains under human control.

Why MCP matters for lead response

The Model Context Protocol creates a common way for AI clients to discover tools and call them on a user's behalf. InstaChime now exposes a scoped MCP endpoint so approved AI clients can inspect recent leads, read analytics, review workflow history, and create controlled test leads without receiving broad account permissions.

Useful MCP-style lead tools

  • List new leads by source, priority, status, owner, or SLA window.
  • Read one lead with original source context and recent lead events.
  • Review lead funnel, source, priority, sentiment, response-bucket, and leaderboard analytics.
  • List workflow definitions and recent workflow runs.
  • Create one controlled test lead when the token has write scope.

Security rules for AI access

AI access should be narrower than normal admin access. InstaChime MCP tokens are workspace-scoped, permission-scoped, revocable, and audited. Sensitive settings are not returned to the AI client. Write access is intentionally limited to controlled test-lead creation until broader mutation tools have explicit approval and stronger review flows.

How to connect an AI client

  1. Open settings and create an API token under API tokens & MCP.
  2. Give the token only the scopes the AI client needs.
  3. Choose Streamable HTTP and configure https://instachime.com/api/mcp with the token as a bearer token.
  4. Initialize with protocol 2025-11-25, then send the initialized notification.
  5. Run tools/list and confirm only expected tools are visible.
  6. Review the MCP setup guide before using the token with real workspace data.

Current protocol and compatibility

InstaChime supports MCP protocol versions 2025-11-25, 2025-06-18, and2025-03-26. The endpoint returns JSON over Streamable HTTP and does not open a standalone SSE stream. Current authentication uses a pre-issued InstaChime bearer token, so the client must support custom Authorization headers.

What the MCP server intentionally cannot do

  • It cannot read leads, analytics, or workflows from another workspace.
  • It cannot reveal notification, CRM, billing, or integration credentials.
  • It cannot send emails, texts, WhatsApp messages, or calls to prospects.
  • It cannot change live lead ownership or delete production records.
  • Its only current write tool creates one controlled test lead when explicitly scoped.

Where to start today

Start with the MCP API-token setup guide, then review quick start, lead routing, CRM lead alerts, and AI lead response.

Start with a controlled test

Connect one scoped AI client.

Create a least-privilege token, inspect the available tools, and test against a controlled lead before using real workspace data.

Start for Free Follow the MCP guide