RLM Tools BSL is an MCP server for 1C BSL codebases that cuts token and context usage far below naive file reading by executing scripts in a sandbox and returning only print() output, deliberately without RAG pre-indexing while offering an optional SQLite index and git grep full-text search for large configurations. The repository is MIT-licensed and not an official 1C project; it is an adaptation of the open-source rlm-tools. Its README cites an April 2026 comparison of six MCP servers claiming 10/10 quality on ten business tasks at the lowest average token use (117K), but those numbers come from the project itself and are self-reported, so treat them with caution. It suits developers and analysts who need deterministic, on-demand analysis of 1C sources; it is not for complete dependency graphs, embedding-based semantic search, or vague business questions. Before adopting, weigh the risks of sandboxed script execution, automatic git grep activation, and optional external LLM calls.
Strongest · Documentation 16/20 Weakest · Security and permissions 9/20
Static review · not runListed 2026-09-10
Read the FMRS scoring method →Runtime：Python 3.10+ · Docker
RLM_UPDATE_INDEX_ON_START
Set to 1 in docker-compose.yml to have the container rebuild registered project indexes at startup; off by default.
RLM_STRATEGY_MODE
Set to full to inline all helper recipes in rlm_start (~20K chars, better for weak models); slim default saves tokens. Set as a service environment variable.
REPOS_ROOT
Mount path for the 1C source root when deploying via Docker; edit it in docker-compose.yml.
Windows (administrator PowerShell): run irm https://raw.githubusercontent.com/Dach-Coin/rlm-tools-bsl/master/simple-install-from-pip.ps1 -OutFile simple-install-from-pip.ps1 then PowerShell -ExecutionPolicy Bypass -File .\simple-install-from-pip.ps1; the script installs from PyPI, registers a Windows service, starts the server and checks health, and re-running it updates to the latest version. Linux: curl -LO https://raw.githubusercontent.com/Dach-Coin/rlm-tools-bsl/master/simple-install-from-pip.sh, then chmod +x simple-install-from-pip.sh && ./simple-install-from-pip.sh. Docker: cp docker-compose.example.yml docker-compose.yml, edit REPOS_ROOT and other variables, then docker compose up -d (not recommended on Windows + Docker Desktop). From source for development: Python 3.10+ and uv; run simple-install.ps1 on Windows or simple-install.sh on Linux. After installation, add an mcpServers entry in .claude.json / mcp.json pointing to http://127.0.0.1:9000/mcp.
{
"mcpServers": {
"rlm-tools-bsl": {
"type": "http",
"url": "http://127.0.0.1:9000/mcp"
}
}
}
Shown for Claude Code. Other clients may use a different file or key (VS Code uses "servers") — the configurator below converts it.
{
"servers": {
"rlm-tools-bsl": {
"type": "http",
"url": "http://127.0.0.1:9000/mcp"
}
}
}
Goes in your project's .vscode/mcp.json (VS Code uses a "servers" key).
claude mcp add --transport http rlm-tools-bsl http://127.0.0.1:9000/mcp
Run it in a terminal; replace any <…> placeholders with your own values first.
The client's MCP tool list should show the six tools: rlm_projects, rlm_index, rlm_start, rlm_help, rlm_execute, rlm_end; the installer also checks the health endpoint at http://127.0.0.1:9000/mcp.
Once connected, you can ask your AI assistant things like:
No matching tools
Listed from the project's documentation, not tested by this site.
RLM Tools BSL is an MCP server for token-efficient analysis of 1C:Enterprise BSL codebases. It is an adaptation of the open-source rlm-tools project to 1C platform specifics: large codebases, source formats (CF/EDT/CFE), metadata structure, Cyrillic code, and XML object descriptions. Its core philosophy is to give an AI assistant codebase search comparable in quality to RAG but without RAG itself — no pre-vectorization — while saving substantial tokens and context. How it works: the agent sends a Python script to the server, the script runs in a sandbox, and only its print() output returns to the agent's context, replacing raw files with compact filtered results. The server exposes six MCP tools (rlm_projects, rlm_index, rlm_start, rlm_help, rlm_execute, rlm_end) and a sandbox with 62 helpers (52 BSL-specific + 8 standard + 2 LLM), plus session caches, timeouts, and security controls. An optional SQLite index of methods and the call graph speeds up large configurations, and a .git directory in the sources automatically enables git_search, a git grep full-text engine. A project registry lets you register sources by name with a required password. Licensed under MIT.
Upstash's official server providing up-to-date third-party library docs for AI coding assistants
AI-powered NPM package analysis MCP server
An independent control layer that gives AI coding agents budgets, brakes, and receipts
GitHub's official MCP server for managing repos, issues, PRs, and workflows via natural language
Source revision b160fa56c506 Data synced 2026-10-19 Read the FMRS scoring method