1C metadata MCP proxy (adapted fork)
1C_MCP_SERVER_OWN
A Python proxy between an MCP client and a 1C HTTP service: reads metadata, object structure and predefined data, with optional OAuth2
Low risk
We rate an entry low when it mostly gives the agent instructions and reference material.
Why this level
- The tools only read metadata and predefined items; writing to the database is not described
- 1C credentials are stored in .env or environment variables on the server
Install
Manual install
python -m src.py_server stdioStdio mode for local clients; set MCP_ONEC_URL, MCP_ONEC_USERNAME and MCP_ONEC_PASSWORD before running.
This is third-party code. Review the repository files before installing.
What it does
The server translates MCP calls into JSON-RPC requests against a 1C HTTP service published from the database as an extension. Four tools are available: listing metadata objects by type and name mask, a detailed object structure with attributes and tabular sections, and listing plus detailed lookup of predefined catalog items. It runs in two modes: stdio for local clients like Claude Desktop and Cursor, and HTTP with /mcp/, /sse, /health and /info endpoints for web scenarios and multiple concurrent users. In HTTP mode, OAuth2 can be enabled with password grant or authorization code plus PKCE, so each client connects with its own 1C credentials instead of one shared account.
Who it is for. For 1C developers who want a ready MCP bridge to a specific database's metadata without writing their own HTTP service from scratch.
Good fit when
- An agent needs to read the list and structure of a specific 1C database's metadata objects
- You need access to predefined catalog items through natural language
- Several users need to connect to the server with their own 1C credentials via OAuth2
Not a fit when
- You need to write data into 1C, not just read metadata and predefined items
- You cannot publish the 1C extension's HTTP service at /hs/mcp/rpc
- You want an independent project rather than an adaptation: the author explicitly credits the source repository vladimir-kharin/1c_mcp
Example request
Show the structure of the Nomenclature catalog with its attributes and tabular sectionsLimitations
The README explicitly calls the project a rework and adaptation of github.com/vladimir-kharin/1c_mcp for the author's own needs, so several architectural decisions are inherited from there. Without OAuth2, all requests run under one configured user. A published 1C extension HTTP service is required on the database side, but the README covers only the Python client and not how to install that extension. The README mentions a testMCP.md file with test results but does not describe the test suite in detail.
How to disable. Stop the python -m src.py_server process and remove the server from your MCP client configuration.
MCP
- Transport
- stdio, http, sse
- Authentication
- API key
| Environment variables | |
|---|---|
| MCP_ONEC_URL required | URL of the 1C database with the published HTTP service. |
| MCP_ONEC_USERNAME required | 1C username, required unless OAuth2 is used. |
| MCP_ONEC_PASSWORD required, secret | 1C user password, required unless OAuth2 is used. |
Security check
- The tools only read metadata and predefined items; writing to the database is not described
- 1C credentials are stored in .env or environment variables on the server
README in short
The Russian README explains the proxy server between MCP clients and a 1C HTTP service, gives a quick start for stdio and HTTP modes, documents four API methods with JSON examples, a configuration parameter table, OAuth2 settings with token timeouts, an architecture diagram and a troubleshooting section. It credits the source project vladimir-kharin/1c_mcp, which this is adapted from, both at the top and the bottom. MIT license.
FAQ
Can this server write data into 1C?
No, the README only describes read tools: listing metadata objects, object structure and predefined items.
Is OAuth2 required?
No, the default auth mode is none and all requests run under the one configured user.
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