1C 8.3 syntax helper MCP
1C Syntax Helper MCP (mcp-1c-helper)
A read-only MCP server searches 1C:Enterprise 8.3 docs in Elasticsearch and returns results via hybrid BM25 plus embedding search
Low risk
We rate an entry low when it mostly gives the agent instructions and reference material.
Why this level
- The server is explicitly read-only and only searches the reference index, writing nothing
- The MCP endpoint is published with no built-in authentication, so access should be network-restricted
Install
Manual install
cp .env.example .env
docker compose up -d --buildStarts the server; Elasticsearch is addressed separately via ELASTICSEARCH_URL.
This is third-party code. Review the repository files before installing.
What it does
The server parses the platform's .hbk help file and indexes it into Elasticsearch, then exposes five tools: search by name, description or keywords via find_1c_help, a quick and a full syntax-element reference, a list of an object's methods and properties, and a context-filtered search over global, object or all. find_1c_help works hybrid: BM25 text search plus kNN over embeddings, merged via RRF, when an embedding service is configured, otherwise search stays purely lexical. A separate script builds a semantic index on top of an already-built lexical one, without touching the original. .hbk indexing is triggered manually; there's no automatic indexing on startup.
Who it is for. For 1C developers who already have, or are willing to run, a shared Elasticsearch and want quick read-only search over the platform's syntax reference right from an agent.
Good fit when
- You want quick search over the 1C platform's syntax reference right from an agent, without opening the syntax helper
- You already have Elasticsearch on the network and can index your platform version's .hbk into it
- You need a list of methods and properties for a specific platform object
Not a fit when
- There's no access to an external Elasticsearch: the server doesn't run one itself, only indexes into and searches an existing one
- You need semantic search but the embedding service is unavailable: without it find_1c_help only works lexically
- You need help for an application configuration, not the platform: the server indexes the platform's own .hbk
Example request
Find what methods the ValueTable object has in the platform reference and show a quick description of the Collapse methodLimitations
Elasticsearch is an external service that must be deployed and addressed separately; the server doesn't run one itself. Semantic search on the help1c_docs_v2 index, by the author's own measurement, gives Recall@3 around 46% and Recall@5 around 50%, meaning it doesn't guarantee the right document lands at the top. The .hbk file with the platform reference isn't shipped in the repository; it must be extracted from your own platform install, and 7z is needed for extraction. Indexing is triggered manually and, without the --reindex flag, only runs if the index is empty.
How to disable. Stop the container with docker compose down and remove the server from your MCP client configuration.
MCP
- Transport
- http
- Authentication
- not required
| Environment variables | |
|---|---|
| ELASTICSEARCH_URL required | Address of the external Elasticsearch holding the reference index. |
| EMBEDDING_URL | Address of the embedding service; empty means search is lexical only. |
| ELASTICSEARCH_INDEX | Index name, defaulting to the semantic help1c_docs_v2 or the older lexical help1c_docs. |
Security check
- The server is explicitly read-only and only searches the reference index, writing nothing
- The MCP endpoint is published with no built-in authentication, so access should be network-restricted
README in short
The README describes the project's composition (a FastMCP server, an .hbk parser and indexer, search, models, config), an environment with an external Elasticsearch and an optional embedding service, hybrid search via RRF with approximate recall metrics, running via docker compose and checking tools/list, a full list of five MCP tools, .hbk indexing commands with the --reindex flag, building a separate semantic index without touching the original, and an example of connecting to opencode as a remote server.
FAQ
Does the server run Elasticsearch itself?
No, the README explicitly treats Elasticsearch as an external service at ELASTICSEARCH_URL; docker compose only brings up the MCP server itself.
Does search work without an embedding service?
Yes, if EMBEDDING_URL isn't set, find_1c_help works purely lexically via BM25, and the other tools already search by exact name.
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