1C Code Search MCP
A local MCP server for semantic search over an exported 1C configuration, with a choice of embedding model and vector database
Low risk
We rate an entry low when it mostly gives the agent instructions and reference material.
Why this level
- Only reads and indexes a local configuration export, changes nothing in the database
- The SSE endpoint can be exposed over the network with no built-in authorization
Install
Manual install
python install.pyInstall dependencies after cloning the repository.
This is third-party code. Review the repository files before installing.
What it does
The server indexes an exported XML 1C configuration and lets you search code by meaning rather than exact text. It supports several infobases at once, each with its own embedding model and vector index. The default model is rubert-tiny2, but any HuggingFace or ONNX-compatible model can be used, and Qdrant, LanceDB or ChromaDB are available as the search backend. A web interface handles configuring the bases and indexing parameters, and the MCP itself is served over SSE, so it can be reached over the network, not just locally.
Who it is for. For 1C developers with large configurations who need an agent to find relevant code by meaning, not just exact text matches.
Good fit when
- The configuration is large and plain grep struggles to find the right code for an agent
- You need to keep several infobases with separate indexes
- You want a choice of local vector databases without an external cloud service
Not a fit when
- The configuration is small and plain file search is enough
- There is no file export of the configuration, only a file-based infobase
- You want a ready cloud service without local setup and indexing
Example request
Find the code in the Trade Management base that handles reserving stock in a warehouseLimitations
The first run downloads and converts an embedding model to ONNX, which needs access to HuggingFace. Indexing is tuned through in-code constants (BATCH_SIZE, EMBED_BATCH_SIZE, CHUNK_SIZE), and changing the chunk size requires a full reindex. The project is small, and the README does not describe behavior under heavy code volumes or load limits. MIT license.
How to disable. Stop the code-search.py process and remove the server from your MCP client configuration.
MCP
- Transport
- sse
- Authentication
- not required
Security check
- Only reads and indexes a local configuration export, changes nothing in the database
- The SSE endpoint can be exposed over the network with no built-in authorization
README in short
The README describes a local Python MCP server for semantic search over 1C code, with a web UI on port 8000. It lists supported embedding models and vector engines, shows a config.yaml example with several bases, explains internal indexing performance constants, and gives ready connection snippets for Claude Desktop, Cursor and Kiro over SSE. MIT license.
FAQ
Can several configurations be connected at once?
Yes, config.yaml supports a list of infobases, each with its own source folder, index and embedding model.
Is Qdrant required?
No, lancedb or chromadb can be chosen instead through the vector_db field.
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