Mcp 1c

What is Mcp 1c?

Mcp 1c is an MCP (Model Context Protocol) server that integrates AI assistants with 1C:Enterprise. It allows AI models to see the metadata of your 1C configuration and generate accurate BSL (1C’s built-in language) code. The server works with any MCP-compatible client, including local models (Ollama, LM Studio), cloud services (Claude, GPT, YandexGPT), and AI-powered IDEs (Cursor, VS Code, JetBrains). It is designed for 1C developers who want to use AI to analyze, query, and modify their live 1C database.

How to use Mcp 1c?

Download the binary for your OS from Releases. Run mcp-1c --install "path-to-base" to automatically set up the required extension in your 1C database. Then start the 1C HTTP service (recommended via Apache/IIS, or use /HTTPPort for quick testing on Windows). Finally, configure your AI client with a JSON block specifying the command and arguments (e.g., --base http://localhost:8080/hs/mcp). The server exposes nine tools for the free version, with fifteen additional tools available in the paid beta.

Key features of Mcp 1c

• One binary, zero dependencies (written in Go).
• 9 tools for live 1C database work.
• Full-text search (smart BM25, regex, exact) with BSL synonym support.
• Sharded indexing (~7 sec for 13,000+ modules) with disk cache.
• Non‑blocking start – server is available immediately.
• Works with any MCP‑compatible client and any AI model.

Use cases of Mcp 1c

• Inspect metadata tree (directories, documents, registers) of your 1C configuration.
• Retrieve object structure (props, tabular sections, dimensions).
• Execute SELECT queries on live data with parameters.
• Search code across all modules with smart ranking or regex.
• Read event log filtered by date, level, and user.
• Get BSL syntax help for built‑in functions and types.

FAQ from Mcp 1c

What is Mcp 1c and what does it integrate with?

Mcp 1c is a local MCP server that connects AI assistants to your 1C:Enterprise database. It integrates with any MCP-compatible client and any AI model (local or cloud), and does not depend on a specific neural network or IDE.

Does Mcp 1c require an internet connection or a cloud AI service?

No. Mcp 1c works with local models (Ollama, LM Studio, llama.cpp) so your 1C code and data never leave your network. The server is a local process that only communicates with your database. Cloud services are optional and are handled by the client, not by Mcp 1c.

How do I install the extension in my 1C database?

Use the --install flag: mcp-1c --install "C:\path\to\base" (Windows) or mcp-1c --install ~/Documents/InfoBase (macOS/Linux). For client‑server databases, add --server --db-user Admin --db-password pass. The server will automatically find the 1C platform, install the extension, and update the database configuration.

What are the system requirements for Mcp 1c?

A binary runs on Windows, macOS, and Linux. The 1C platform version 8.3.x or 8.5.x (commercial or eductional) is required. HTTP access to a 1C web service (via Apache/IIS or /HTTPPort on Windows) is needed. macOS does not support the 1C HTTP service natively; a Windows VM is recommended.

How do I configure my AI client to use Mcp 1c?

Add the following JSON to your MCP client settings:

{ "mcpServers": { "1c": { "command": "/path/to/mcp-1c", "args": ["--base", "http://localhost:8080/hs/mcp"] } } }

On Windows, use double backslashes in the binary path. This configuration works identically for all clients (Claude Desktop, Ollama, LM Studio, Cursor, VS Code, etc.).