1C Context MCP
An MCP server that unpacks 1C's .hbk help, indexes it in Qdrant and gives the agent search over API, snippets and standards
Low risk
We rate an entry low when it mostly gives the agent instructions and reference material.
Why this level
- Reads and indexes help and configuration metadata, never writes into the 1C database
- The index and metadata exports stay inside your own Docker infrastructure
Install
Manual install
make quick-start-prebuiltDownloads a ready public Qdrant/BM25 index and brings up MCP at http://localhost:8050/mcp with no local .hbk help.
This is third-party code. Review the repository files before installing.
What it does
The project parses 1C platform help in .hbk format into structured JSONL, builds a vector index in Qdrant from it and runs an MCP server for searching API references, code snippets, development standards and configuration metadata. It runs as several Docker Compose services: mcp serves MCP over HTTP, qdrant holds the indexes, redis holds the ingest cache, and ingest-worker builds the index and runs a scheduled watchdog. You can bring up a ready public index with no local 1C help via make quick-start-prebuilt, or build your own index from your own .hbk and configuration metadata export.
Who it is for. For 1C developers who want Cursor or another agent to surface platform API, ready snippets and standards directly from their configuration's context.
Good fit when
- You need fast search over 1C platform help right inside the agent instead of switching to web docs
- You have your own .hbk help and configuration metadata worth indexing
- You need MCP over streamable HTTP with ready Cursor configuration examples
Not a fit when
- You do not want to run a multi-service Docker Compose setup just for an MCP server
- You have no .hbk help of your own and the public demo index does not fit your data (the author forbids publishing backups built from private or NDA configurations)
- You need a lightweight search without Qdrant and Redis
Example request
Find the method signature for long-running operations in 1C help and show a ready call snippetLimitations
Full operation needs Docker and several services: mcp, qdrant, redis, ingest-worker. The public demo index on Mail.ru Cloud is for evaluation only, not production. Your own index needs 1C platform .hbk help and a metadata export via the tools/1c/MetadataExport.epf processing. The project changes actively, and its license is not machine-readable via the GitHub API.
How to disable. Stop the stack with docker compose down and remove the server from your MCP client settings.
MCP
- Transport
- http
- Authentication
- not required
| Environment variables | |
|---|---|
| HOST_HELP_SOURCE_BASE | Path to local 1C platform .hbk help for building your own index |
Security check
- Reads and indexes help and configuration metadata, never writes into the 1C database
- The index and metadata exports stay inside your own Docker infrastructure
README in short
The bilingual README describes a four-service pipeline and two start paths: a quick one with a public prebuilt index or a full one with local 1C help. It shows make commands for running, ingesting and reindexing with cleanup, the streamable HTTP endpoint and a Cursor config example. The repository has a separate docs/cursor-examples folder with ready SKILL.md files and rules for different ways of using this MCP with Cursor.
FAQ
Do I have to build my own index?
No, make quick-start-prebuilt downloads and restores a public demo index with no local .hbk help needed.
How often does a self-built index refresh?
The watchdog runs on startup and every 10 minutes, a full ingest runs daily at 3:00, or you can trigger it manually with make ingest.
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