ROCTUP/1c-mcp-toolkit

What it solves

1C MCP Toolkit provides a way to connect AI agents (such as Claude or Kiro) to 1C:Enterprise databases. It eliminates the need for complex configuration changes, server publication, or COM connections, allowing AI agents to interact with 1C data and logic directly.

How it works

The project offers two primary modes of operation:

1. Embedded Server: An HTTP server runs directly within a 1C processing file (.epf), requiring no Python installation on the host machine.
2. Proxy Mode: A Python-based server (FastAPI + MCP SDK) acts as an intermediary. The 1C processing file connects to this proxy via HTTP Long Polling, and the AI agent communicates with the proxy via the Model Context Protocol (MCP) or a REST API.

Both modes allow the AI agent to execute queries, run arbitrary 1C code, and access metadata and event logs.

Who it’s for

Developers and administrators of 1C:Enterprise systems who want to integrate AI agents into their business processes for data retrieval, automation, and system analysis.

Highlights

• Dual Architecture: Supports both a lightweight embedded server and a scalable Docker-ready proxy server.
• Comprehensive Toolset: Provides tools for executing 1C queries, running BSL code, fetching metadata, and even taking screenshots of the active 1C window (Windows only).
• Data Anonymization: Built-in support for replacing sensitive personal or confidential data with stable tokens to maintain privacy before sending data to an AI agent.
• Flexible Integration: Compatible with the Model Context Protocol (MCP) for native AI agent integration and a standard REST API for agents that do not support MCP.
• Broad Compatibility: Works with 1C:Enterprise versions 8.2.13+ and 8.3.25.