v8help: CLI and MCP for 1C help with hybrid search
v8help
An MCP server and CLI for 1C that build full-text and vector indexes over the platform's .hbk help and parse PascalCase Russian identifiers
Low risk
We rate an entry low when it mostly gives the agent instructions and reference material.
Why this level
- Only reads the local platform help file and builds a search index
- With vector search, query text may be sent to the configured embedder
Install
Manual install
python -m venv .venv && pip install -e .Installing registers two console scripts: v8help and v8help-mcp.
This is third-party code. Review the repository files before installing.
What it does
The tool extracts HTML pages from the 1C platform's help container, converts them to Markdown while rewriting internal v8help://... links to relative ones, and builds a SQLite FTS5 full-text index with lexical expansion that can split identifiers like СтрНайтиПоРегулярномуВыражению into their component words. Ranking weighs fields: a match in the title or the Description section outweighs one in the article body. Vector and hybrid search are supported through an OpenAI-compatible embeddings API, with long articles chunked and neighbor metadata attached. Index building is asynchronous: build returns a job_id immediately, progress is tracked via build_status, and search is not blocked while it runs.
Who it is for. For 1C developers who need precise search over platform help that understands Russian PascalCase identifiers, with optional meaning-based search.
Good fit when
- You need 1C help search that correctly splits compound Russian identifiers
- You want both a terminal CLI and an MCP server for an agent from one install
- You have ready indexes without an installed platform: they can be reused per the docs instead of rebuilding
Not a fit when
- You need vector search only, without FTS5: the tool is hybrid by default
- You have neither an installed platform nor a ready index: there is nothing to search without one of them
Example request
Use v8help to find the СтрНайтиПоРегулярномуВыражению method and show its parametersLimitations
Rebuilding the index needs an installed 1C:Enterprise platform with a bin folder containing .hbk files; search alone needs just a ready database. Vector or hybrid search needs a configured embedder: LM Studio, Ollama or Hugging Face. A Docker option is documented separately, with the database on a volume and optional index auto-loading.
How to disable. Remove the v8help entry from your MCP client config and, if needed, uninstall the v8help and v8help-mcp console scripts.
MCP
- Transport
- stdio, http
- Authentication
- not required
Security check
- Only reads the local platform help file and builds a search index
- With vector search, query text may be sent to the configured embedder
README in short
The README describes an MCP tool and CLI for reading, indexing and searching 1C help in .hbk format. It gives a quick start with copying the config, the build and search commands, and links to separate docs on configuration, embeddings, Docker deployment and development. It separately credits the HTML-to-Markdown converter ported from the hbk-to-md project.
FAQ
Is the 1C platform needed for every search?
No, the platform is needed only to build the index; search works from an already built database, and ready indexes exist without an installed platform.
How is the article title weighted in ranking?
The title, description and body fields are weighted 9 to 3 to 1, so a match in the title or method description matters more than one in the body.
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