1c-nexus-infra: code graph, RAG and MCP bridges for 1C
1c-nexus-infra
Standalone infrastructure for a 1C developer: a static code call graph, a RAG-style decision base with a lifecycle, and bridges to external MCP servers
Medium risk
We rate an entry medium when the tool runs code, makes network calls or reads project files. Check what exactly it does before installing.
Why this level
- BSL candidate evaluation sends code fragments to an external LLM endpoint when the gateway is enabled
- The MCP bridges give access to the external RSV Data and CodexTestBridge services
Install
Manual install
export NEXUS_WORKSPACE_ROOT="$(pwd)" && python3 -m pip install -e .Installs the package with the workspace root defaulting to the current directory.
This is third-party code. Review the repository files before installing.
What it does
The project bundles several engineering practices into one portable framework for 1C development. Graph V2 builds a static code navigation graph and answers call and counter-control queries. RAG FSM 2.0 keeps a knowledge_base.json decision store with a single writer under a file lock and lifecycle statuses of under_review, unstable, success, failed, partial. The Prompt Cache Contract keeps a byte-identical static prompt prefix simultaneously across .clinerules, .cursorrules, AGENTS.md and system_prompt.md, with its own integrity check. Best-of-N BSL evaluates BSL code candidates with fallback roles through a separate LLM runner. MCP bridges connect the infrastructure to an external RSV Data MCP and a CodexTestBridge.
Who it is for. For experienced 1C developers who need infrastructure around the agent: a code graph, a lifecycle-tracked decision base, and BSL code candidate scoring, rather than a single point tool.
Good fit when
- You need a static call graph over a 1C codebase without hitting the model
- You need a structured project decision base that survives agent session changes
- You need to score several BSL code candidates instead of trusting the first answer
Not a fit when
- You need one simple MCP server, not a multi-component infrastructure stack
- You are not ready to learn the project's own terminology: the OMVL prefix, the WAN25 gateway, RAG FSM
Example request
Build a dependency graph for this common module with map_workspace_graph_v2 and show who calls itLimitations
The README is written in dense technical language full of internal terminology and duplicated English and Russian sections, making it hard to get into quickly. Best-of-N BSL and the shared WAN25 gateway expect a separate OpenAI-compatible endpoint with its own rate limit. Secrets are set only in .env, not in git; the repository carries only an example with placeholders.
How to disable. Remove the MCP bridges from the agent's config and, if the infrastructure is no longer needed, delete the project folder entirely.
MCP
- Transport
- http
- Authentication
- API key
| Environment variables | |
|---|---|
| NEXUS_WORKSPACE_ROOT | The framework root, defaulting to the module directory. |
| RSVDATA_MCP_URL | Address of the external RSV Data MCP server. |
| WAN25_API_KEY secret | Secret for the shared WAN25 gateway, when enabled. |
Security check
- BSL candidate evaluation sends code fragments to an external LLM endpoint when the gateway is enabled
- The MCP bridges give access to the external RSV Data and CodexTestBridge services
README in short
The README lists the system's parts in a table: Graph V2, RAG FSM 2.0, the Prompt Cache Contract, Best-of-N BSL, MCP bridges, and workspace hygiene utilities. It gives a quick start with the NEXUS_WORKSPACE_ROOT variable, installing the package and checking prompt contracts, a table of environment variables with purpose and defaults, and verification instructions via pytest and py_compile.
FAQ
What is the OMVL prefix?
A byte-identical static prompt prefix, OMVL_STATIC_PREFIX_V1, repeated across several config files and checked by a dedicated script for a match.
Is the WAN25 gateway required?
No, WAN25_GATEWAY_ENABLED is off by default, keeping role-specific routes instead of one shared gateway.
Related
An MCP server built into the Netdata agent: metrics, logs, alerts and live process, service and container data for an AI assistant
GitHub's official MCP server: code, issues, pull requests, Actions and security alerts straight from the agent
Agent Skills for Google products
Agent Skills for Google products and technologies
Official Google skill collection for working with Google Cloud, BigQuery, GKE, ads and analytics from an agent
AWS's official MCP server suite: docs, IaC, containers, serverless, databases, cost and monitoring