1C AI Assistant MCP proxy
1c-ai-mcp
An MCP server that forwards questions, code review and code explanation requests to the 1C AI Assistant cloud service
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- Sends source code and questions to the external code.1c.ai cloud service
- The access token is passed in the agent config as a command-line argument
Install
Manual install
{
"mcpServers": {
"1c-ai-mcp": {
"command": "node",
"args": ["/путь/к/1c-ai-mcp.js", "--mcp", "--token=<ваш-токен>"],
"env": { "AI_PROJECT_ROOT": "/путь/к/проекту" }
}
}
}The .mcp.json snippet from the README.
This is third-party code. Review the repository files before installing.
What it does
The server acts as a proxy to the code.1c.ai cloud service, known as the 1C AI Assistant (1С:Напарник), and gives an agent five tools: ask a 1C development question, review code for errors and performance issues, fix found errors, explain what a piece of code does, and fetch the continuation of an answer if a call did not finish within 30 seconds. The proxy runs as a regular stdio MCP server via node with the --mcp flag and a code.1c.ai access token, and the files the model can see can be limited to a project root with AI_PROJECT_ROOT.
Who it is for. For 1C developers with code.1c.ai access who want to call the 1C AI Assistant from any MCP-compatible agent, not only its own client.
Good fit when
- You want to ask a 1C development question directly from Claude Code, opencode or another MCP client
- You need to review BSL code for errors and performance issues before a review
- You need an explanation of unfamiliar or legacy BSL code
Not a fit when
- You do not have a code.1c.ai access token
- You need to work with a live 1C database: the tools do not read or change the infobase, only code and questions
Example request
Review this module for errors and explain what the РассчитатьСкидку function doesLimitations
You need your own code.1c.ai token, either paid or trial, which the project does not provide. Requires Node.js 22+. A tool call is capped at 30 seconds, so longer answers need a separate GetResult call. A single-author project with no stated support policy.
How to disable. Remove the 1c-ai-mcp entry from .mcp.json or your agent's MCP configuration.
MCP
- Transport
- stdio
- Authentication
- API key
| Environment variables | |
|---|---|
| AI_TOKEN required, secret | The code.1c.ai access token, an alternative to the --token flag. |
| AI_PROJECT_ROOT | The project root, controlling which files the model can see. |
| AI_FULL_LOG | Enables tracing of requests and responses. |
| AI_SESSION_FILE | Path to the session cache file. |
| AI_TRACE_LOG | Path to the trace log file when AI_FULL_LOG is enabled. |
Security check
- Sends source code and questions to the external code.1c.ai cloud service
- The access token is passed in the agent config as a command-line argument
README in short
The README lists the proxy's five tools, requirements (Node.js 22+, a code.1c.ai token), two install options: download a release zip or build from source. It gives CLI parameters and environment variables with a table of requirements and defaults, plus ready config snippets for opencode and Claude Code. MIT license.
FAQ
Where do I get a token?
The token comes from the code.1c.ai (1C AI Assistant) service; 1c-ai-mcp only proxies requests to it and does not issue tokens.
What if a call does not finish within 30 seconds?
Call the GetResult tool, which returns the continuation of the same answer.
Related
A skills library that gives coding agents a development process: brainstorming, planning, TDD, subagents and code review
Skills for real engineers by Matt Pocock
Skills For Real Engineers
Small composable skills for engineering with agents: plan grilling, TDD, bug diagnosis, code review and architecture
GitHub toolkit for spec-driven development: the specify CLI adds agent commands and skills to a project, from principles to implementation
Reference MCP servers
Model Context Protocol servers
Official reference MCP servers: Filesystem, Fetch, Git, Memory, Sequential Thinking, Time and Everything