DaJet MCP Server
A DaJet Metadata based MCP server that exposes 1C metadata structure and runs SELECT queries against the database
High risk
We rate an entry high when the tool writes to external systems, handles money, production databases or secrets, or runs arbitrary commands. The CLI installs it only with your consent.
Why this level
- execute_query runs agent-composed queries against a real database
- Database connection credentials are stored in an unencrypted config file
Install
Manual install
docker pull zhichkin/dajet-mcp-server && docker run --name dajet-mcp-server --user=root -it -p 5000:5000 zhichkin/dajet-mcp-serverRuns in Docker on Windows or Linux.
This is third-party code. Review the repository files before installing.
What it does
The server is written in C# and built on the DaJet Metadata library; it reads 1C:Enterprise 8 metadata together with the real table and field names in the database. It is configured through datasources.json, which lists connections to several databases at once, and an agent addresses each by name. MCP tools: listing metadata type names, listing databases, resetting the cache, describing a database, getting a database's full metadata structure, searching objects by a name fragment, getting an object by type and name or by numeric code, resolving UUIDs to object names, and execute_query, which runs a parameterized SELECT in DaJet Script and returns the result as JSON. It installs from a binary release for .NET 10 or as a Docker Hub image, and runs on Windows and Linux.
Who it is for. For 1C developers and integrators who need an agent to understand the real table structure of a specific database and read data directly with SQL-like queries.
Good fit when
- You need an agent to see actual DBMS table and field names, not just 1C metadata objects
- You need agent access to several 1C databases at once by source name
- Read-only access via SELECT is enough, without writing to the database
Not a fit when
- Arbitrary agent-issued SELECT queries against a production database are not acceptable
- You have neither .NET 10 nor the ability to use Docker
- You need to write data, not just read it and read metadata
Example request
Find the Nomenclature catalog structure in the MS_TEST database and select the first record with code 333Limitations
execute_query runs a query that the agent itself composes, and the README does not describe a hard limit on result size. Database connection strings, including credentials, are stored in the plain datasources.json file. The README only shows connecting the MCP server in Cursor; the default HTTP address and port come from appsettings.json.
How to disable. Stop the DaJet.Mcp.Server.exe process or the container with docker rm -f dajet-mcp-server, and remove the server from your client's mcp.json.
MCP
- Transport
- http
- Authentication
- not required
Security check
- execute_query runs agent-composed queries against a real database
- Database connection credentials are stored in an unencrypted config file
README in short
The README describes installing via a .NET 10 binary release or Docker Hub, configuring the server address in appsettings.json and data sources in datasources.json, a sample mcp.json config for Cursor, a table of nine MCP tools with parameters, an execute_query example in DaJet Script with parameters, and supported parameter data types. It states the functionality is based on the DaJet Metadata library and is similar to DaJet HTTP Server.
FAQ
Can several databases be connected at once?
Yes, datasources.json supports a list of sources, and an agent addresses the one it needs by name.
Can execute_query change data?
No, the README describes it as running an arbitrary parameterized SELECT, meaning read-only.
Related
166 skills for scientific work: bioinformatics, cheminformatics, clinical data, geospatial analysis and 100+ databases
Google's open-source MCP server for databases: ready tools for Postgres, MySQL, BigQuery, Spanner and more, plus custom tools in tools.yaml
Official Hugging Face skills: Hub operations via the hf CLI, datasets, model training, Spaces, evals and deployment
A visualization language for agents and an MCP server: neat charts from a simple semantic spec