1C-ai-mcp: MCP bridge to 1C:Napparnik
1C-ai-mcp
An MCP server that exposes the cloud 1C:Napparnik (1C.ai) assistant and a local BSL Language Server to agents in Cursor and other IDEs
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
- rename_symbol and format_bsl_code change the project's source code
- The ONEC_AI_TOKEN and code content are sent to the external 1С.ai cloud service
Install
Manual install
git clone https://github.com/novoros-id/1C-ai-mcp.git && cd 1C-ai-mcp && pip install -r requirements.txtAfter installing dependencies, create a .env file with ONEC_AI_TOKEN.
This is third-party code. Review the repository files before installing.
What it does
The server wraps the cloud 1C:Napparnik (1C.ai) service into ask_1C_expert and explain_1C_syntax MCP tools for platform questions, and embeds a local bsl-language-server as audit_code_expert, check_bsl_standards, format_bsl_code, get_code_completions and rename_symbol for static analysis, formatting and symbol renaming in BSL code. Separately there is pack_project and analyze_project to pack a project into context via Repomix, git_analyze_diff and git_conflict_helper for commits and merges, and debug_version_compare and generate_method_docs. On startup the server detects Java and Node.js itself; missing tools (audit, packaging) are disabled automatically, and the BSL Language Server can update in the background from GitHub. The cloud tools need an ONEC_AI_TOKEN obtained at code.1c.ai through a 1C:ITS account.
Who it is for. For 1C developers in Cursor or another VS Code based IDE who want cloud 1C:Napparnik hints combined with BSL static analysis in one MCP server.
Good fit when
- You already have access to 1C:Napparnik and an ONEC_AI_TOKEN
- You need hybrid code auditing: BSL Language Server plus AI explanations
- You need BSL formatting, completion and safe refactoring right through MCP
Not a fit when
- You have no 1C:ITS account to obtain a 1C.ai token
- You cannot install Java and Node.js in the environment: some tools will be disabled
- You need a fully local server with no calls to the 1C cloud service
Example request
Run a hybrid audit on the OбщийМодуль1.bsl module and explain what is wrong with the reported standard violationsLimitations
The key ask_1C_expert and explain_1C_syntax tools depend on the cloud 1C.ai service and require a paid or trial token tied to a 1C:ITS account. Auditing via BSL Language Server requires Java 11+, and project packing via Repomix requires Node.js and npm; without them the corresponding tools are disabled without crashing the server. rename_symbol changes source code and its references, so results should be checked before committing.
How to disable. Stop the MCP server, remove it from your MCP client config, and delete the ONEC_AI_TOKEN from .env.
MCP
- Transport
- stdio
- Authentication
- API key
| Environment variables | |
|---|---|
| ONEC_AI_TOKEN required, secret | Access token for the cloud 1C.ai service, issued in the code.1c.ai control panel via a 1C:ITS account. |
| CODEBASE_OUTPUT_NAME | The file name the Codebase Manager packs the project into, an optional setting. |
Security check
- rename_symbol and format_bsl_code change the project's source code
- The ONEC_AI_TOKEN and code content are sent to the external 1С.ai cloud service
README in short
The README describes four functional layers in detail (1C:Napparnik, the BSL analyzer, the Codebase Manager, git integration), technical solutions like the Triple-Nest strategy for the 1C.ai API and SSE parsing, installation via pip and auto_setup.py, obtaining a token at code.1c.ai, system requirements (Java 11+, Node.js, Python 3.10+) and a full list of ten MCP tools with descriptions.
FAQ
What if Java or Node.js are missing?
The server keeps running, but the audit and project-packing tools turn off automatically.
Where do you get an ONEC_AI_TOKEN?
At code.1c.ai after signing in with a 1C:ITS account, from the control panel.
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