MCP BSL Platform Help Context
An MCP server for the 1C platform API docs read straight from shcntx_ru.hbk help files, with keyword, semantic and hybrid search
Low risk
We rate an entry low when it mostly gives the agent instructions and reference material.
Why this level
- Only reads platform reference files and serves documentation for reading
- Does not connect to a 1C database and does not execute code
Install
Manual install
pip install -e ".[local]"Installs the package along with the dependencies for semantic and hybrid search (sentence-transformers, torch).
This is third-party code. Review the repository files before installing.
What it does
The server reads the binary shcntx_ru.hbk platform help file from a 1C:Enterprise install (or uses pre-exported JSON) and gives an agent nine tools: API search in three modes (keyword, semantic, hybrid), a detailed method or property card by exact name, a type's member list, constructor signatures, platform version info, BSL coding style guidance, and strict-typing documentation with full-text search over it. It supports several platform versions at once with automatic selection of the nearest one, three transports (stdio, SSE, Streamable HTTP), and both Russian and English API names. It is a Python port of the original Kotlin mcp-bsl-platform-context project, with its own independent test coverage (359 tests).
Who it is for. For 1C developers in Claude Desktop, Cursor or another MCP client who need precise platform API answers instead of the model's approximate memory.
Good fit when
- You need the exact signature of a 1C platform method or property with parameters and types
- You need natural-language search about how to do something in BSL
- You work with several 1C platform versions at once
Not a fit when
- You have no access to an installed 1C platform with shcntx_ru.hbk and no pre-exported JSON
- You need help for an application configuration, not the platform itself
- You need semantic or hybrid search on the CPU Docker image without an external embedding API: only keyword works by default there
Example request
Show me the signature of the Найти method on ТаблицаЗначений and explain the Колонки parameterLimitations
The CPU Docker image ships without local ML models and defaults to keyword mode only; semantic and hybrid search need an external OpenAI-compatible API. The first run with a new embedding model downloads it from HuggingFace. The project depends on the binary HBK format, with platform version support limited to the explicitly stated range (up to 8.3.27 and newer).
How to disable. Stop the mcp-bsl-context process or Docker container and remove the bsl-context server from your MCP client config.
MCP
- Transport
- stdio, sse, http
- Authentication
- not required
| Environment variables | |
|---|---|
| MCP_BSL_PLATFORM_PATH required | Path to the 1C platform install folder containing shcntx_ru.hbk, required for the hbk data source. |
| MCP_BSL_SEARCH_DEFAULT_MODE | Default search mode: keyword, semantic or hybrid. |
Security check
- Only reads platform reference files and serves documentation for reading
- Does not connect to a 1C database and does not execute code
README in short
The README lists capabilities (three search modes, multi-version support, YAML config, bilingual support), tables of MCP tools for search and documentation, a description of the three keyword search strategies and the semantic RAG pipeline, pip installation with extras, a full list of environment variables and CLI flags, Claude Desktop and Cursor integration, running in Docker with CPU and GPU profiles, the package's layered architecture, and instructions for AI assistants on tool call order.
FAQ
Is a local ML model required?
No, the base pip install -e . gives only keyword search with no ML dependencies; semantic and hybrid need the [local] extras or an external API.
Can it be used without an installed 1C platform?
Yes, via the json data source with files prepared in advance by platform-context-exporter.
Related
Microsoft's utility and MCP server that turn PDF, Word, Excel, PowerPoint, HTML and other files into Markdown for models
Anthropic skills
Skills
Anthropic's official skill examples: docx, pdf, pptx and xlsx documents, design, MCP builder and the Agent Skills spec
A skill that builds editable PPTX decks from PDF, DOCX, URLs and Markdown with native shapes, charts and animations
A skill that draws editorial diagrams as standalone HTML and SVG files: architecture, flowcharts, ER, funnels and about forty more types