rlm-tools-bsl MCP
An MCP server for analyzing large 1C codebases with less context: the agent runs scripts server-side and gets only a compact result
- 218 ★ on the repository
- Last commit
- MIT
Extended access
Why this level
- Runs agent Python code in a sandbox: a separate process per session, an AST gate, memory limits. The author notes it is not an OS security boundary
- Installs as a system service; a read-only user or Docker removes the remaining risk
- If configured, llm_query helpers send code fragments to an external model (OpenAI-compatible or Anthropic); a local Ollama also works
Install
- Works in
- Services
Manual install
curl -LO https://raw.githubusercontent.com/Dach-Coin/rlm-tools-bsl/master/simple-install-from-pip.sh
chmod +x simple-install-from-pip.sh && ./simple-install-from-pip.shLinux: installs the PyPI package and registers a systemd service.
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- Runs agent Python code in a sandbox: a separate process per session, an AST gate, memory limits. The author notes it is not an OS security boundary
- Installs as a system service; a read-only user or Docker removes the remaining risk
- If configured, llm_query helpers send code fragments to an external model (OpenAI-compatible or Anthropic); a local Ollama also works
This is third-party code. Review the repository files before installing.
What it does
It adapts rlm-tools to 1C: instead of reading whole files, the agent sends a short Python script that runs in a sandbox next to the sources, and only the printed result reaches the context. The sandbox has BSL helpers: list module procedures, read one procedure body, build call graphs, find references to a metadata object, expand defined types. Designer and EDT dumps are supported. An optional SQLite index speeds up large configurations, and with git present, full-text search via git grep is enabled.
Who it is for. For 1C developers, team leads and analysts who explore large unfamiliar configurations.
Good fit when
- You need to understand an unfamiliar subsystem quickly without setting up RAG
- The agent's context fills up reading ERP modules
- You need daily review of fresh code from the repository
Not a fit when
- You need a full dependency graph of the whole configuration
- You need semantic search over object descriptions
Example request
Figure out how the АвансовыйОтчет document posts and describe its register recordsMCP
- Transport
- http
- Authentication
- not required
Environment variables
RLM_STRATEGY_MODE- Session start mode: slim by default or full
RLM_UPDATE_INDEX_ON_START- Rebuild indexes on container start
Limitations
Works on sources on disk and does not read the live base. On slow disks and ERP-sized configurations without a prebuilt index, timeouts are likely. The llm_query helpers need an OpenAI-compatible or Anthropic API key; everything else works without them. Docker Desktop on Windows is much slower, so install on the host.
How to disable. Stop the rlm-tools-bsl service and remove the server from the client config.
FAQ
Why a password for project registration?
It guards index management and registry changes so the agent cannot run them without human confirmation.
Can it be used alongside RAG?
Yes, the README covers combined use: RLM for targeted analysis, RAG for fuzzy questions.
Discussion
No questions or reviews yet.
Related
All 1C:Enterprise toolsMCP server with the 1C platform reference and validation of BSL code against the real API of a chosen platform version
A Dockerized MCP server for searching 1C dumps: structural index, call graph and semantic code search
An MCP server for Claude Code: full-text search over multiple 1C XML+BSL configuration exports in a local SQLite FTS5 database
A local OneScript MCP server for searching a 1C configuration dump right from Cursor
A set of skills, rules and MCP servers for 1C Enterprise 8.3 development where the agent works from real code and docs, not from memory
A single-binary Go MCP server: the agent sees 1C metadata, searches code, runs read-only queries and writes accurate BSL
README in short
The README opens with installation: scripts for Windows with a service and Linux with systemd, Docker and building from source, then the client config. It explains RAG versus RLM, lists scenarios where the tool fits and where it does not, and compares several MCP servers on one configuration. It covers the project registry, indexing, session start modes and a ready agent instruction. MIT licensed.