1C Metacode MCP Server
A Neo4j graph of 1C configuration metadata and code with MCP tools, a web console and a built-in search agent
Install
docker compose up -dBefore starting, prepare .env and docker-compose.yml from the examples and place the dump under data/prj1.
This is third-party code. Review the repository files before installing.
What it does
The server loads configuration metadata into Neo4j from a configuration report or XML dump, along with extensions, forms, roles, event subscriptions and predefined values. From modules it takes routine signatures and bodies, builds a call graph and offers full-text, vector and hybrid code search. Objects are linked through attributes, document register movements and access rights, and LLM summaries can be generated per object. Agents reach this over MCP and people through a web console; data reloads incrementally on a schedule.
Who it is for. For 1C teams maintaining large configurations with extensions who want to give an agent a full map of relationships.
Good fit when
- You need to see which objects a change to an attribute or register affects
- The configuration has several extensions and you need to compare their objects with the base
- You need to find routines by meaning, not only by name
Not a fit when
- You cannot run Docker with Neo4j on a machine or server
- Plain search over the dump is enough without a graph database
Example request
Use 1c-metacode to find all documents that post to the counterparty settlements register and their posting handlersLimitations
Requires Docker, Docker Compose and free ports for Neo4j and the server. Semantic search and object summaries need an embeddings or LLM provider, meaning keys and possibly a cloud API whose availability from Russia depends on the provider. AGPL-3.0 licensed.
How to disable. Remove the 1c-metacode block from the client MCP config and stop the containers with docker compose down.
MCP
- Transport
- http, sse
- Authentication
- not required
| Environment variables | |
|---|---|
| NEO4J_PASSWORD required, secret | Password for the neo4j user |
| PROJECT_NAME required | Project name matching the data folder |
| WEB_CONSOLE_ENABLED | Enables the web console |
| MCP_USE_SSE | Switches the transport to SSE |
Security check
- Starts Neo4j and an HTTP server whose MCP endpoint has no built-in auth
- Can send code and metadata to an external embeddings or LLM provider
README in short
The README lists what gets loaded and the graph model with a node and relationship diagram. It runs through Docker Compose with Neo4j, exposing Neo4j Browser, an MCP server over streamable HTTP or SSE, and a token-protected web console. The tool reference, search modes, extensions and the built-in agent are documented separately in docs/. AGPL-3.0 licensed.
FAQ
Where does metadata come from?
From the text configuration report or directly from the XML dump; code comes from the XML dump.
Can it hold several infobases?
Yes, it supports multiple projects, each made of a base configuration plus extensions, with search filtered per project.
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