Kommo MCP Server is a Model Context Protocol (MCP) server that gives AI agents (Claude Code, Claude Desktop, Cursor, Gemini CLI, Qwen Code and other MCP clients) ready-made tools for Kommo CRM data — the international version of amoCRM: duplicate detection, manager performance, pipeline and traffic-source analytics.
Built by api-master.ru. Access is granted by request.
We implement and configure amoCRM and Kommo for sales teams. That experience — hundreds of pipelines, custom fields and integrations — is packaged into MCP functions: ready-made methods the agent calls directly instead of re-learning the Kommo API and your data structure every time.
What it covers:
claude mcp add --transport http kommo <server_url> \
--header "Authorization: Bearer <your_token>"{
"mcpServers": {
"kommo": {
"httpUrl": "<server_url>",
"headers": {
"Authorization": "Bearer <your_token>"
}
}
}
}{
"mcpServers": {
"kommo": {
"url": "<server_url>",
"headers": {
"Authorization": "Bearer <your_token>"
}
}
}
}It is a server built on the open Model Context Protocol (MCP) that exposes Kommo CRM data — deals, contacts, pipelines, tasks, users — as ready-to-call tools for an AI agent. The agent gets pre-computed metrics instead of learning the Kommo API and your account field structure.
Request access from the developer. You receive a server URL and a token scoped to your Kommo account, then add it with: claude mcp add --transport http kommo <server_url> --header "Authorization: Bearer <token>".
Any client that supports remote MCP servers: Claude Code, Claude Desktop, Cursor, Gemini CLI, Qwen Code and others.
The agent compares records via MCP tools on key fields (phone, email, company) and surfaces likely duplicates for manual or automatic merging.
From Kommo’s own data — response time, pipeline-stage movement, task timeliness. The logic is standardized, so different agents and reports never disagree.
Access is granted per request with a token scoped to one account; the MCP tools are limited to analytical and read-oriented operations.
Yes — consistent metric logic across every client account is one of the core use cases.
Pricing depends on data volume and scope — message the developer on Telegram @mislawsky.
Updated: