Semantic search over Jivo and SaluteBot docs
MCP Docs Hub
A collection of MCP servers with FAISS indexes for semantic search over Jivo, SaluteBot and SberSalute docs: search, file reads and contents
Low risk
We rate an entry low when it mostly gives the agent instructions and reference material.
Why this level
- It only reads local documentation and the index, with no writes to external systems
Install
Manual install
cd jivo-docs-mcp && python -m venv venv && venv/bin/activate && pip install -r requirements.txt && python builder.pyExample for jivo-docs-mcp; the other servers follow the same steps in their own folder.
This is third-party code. Review the repository files before installing.
What it does
The repository holds three uniform FastMCP servers, each with its own Markdown documentation set and a FAISS vector index: jivo-docs-mcp for Jivo's chat widget API, webhooks, Chat API and Bot API, salutebot-docs-mcp for the SaluteBot chatbot platform (147 doc files), and sber-code-docs-mcp for SberSalute Code covering SmartApp, JS API and NLP (179 doc files). Each server exposes seven tools: semantic search via LlamaIndex and the paraphrase-multilingual-MiniLM-L12-v2 embedding model, exact text search, reading a full file, listing categories and files, finding a file by name, and getting a table of contents.
Who it is for. For developers building bots and SmartApps on SaluteBot and SberSalute Code who want fast semantic search over large documentation sets right from an agent.
Good fit when
- You need to quickly find the right SaluteBot or SberSalute Code doc section by meaning, not an exact match
- There is a lot of documentation (147-179 files) and searching it manually is slow
- You want exact text search as a fallback to semantic search
Not a fit when
- You need current documentation: the snapshot is dated November 30, 2025, and the docs themselves are not bundled, only rebuilt by the indexer
- You are not ready to build the FAISS index yourself: no ready index files ship in the repository
Example request
Find in the SaluteBot documentation how to set up named entities for a botLimitations
Index files (faiss_index.bin, storage/) are not part of the repository; you build them yourself with the builder.py script after cloning, which can take time and needs network access to download the embedding model. The documentation is dated to a specific check (November 30, 2025) and may lag behind current docs. Future plans are stated, but the README gives no concrete timeline for new platforms.
How to disable. Remove the relevant server (jivo-docs, salutebot-docs) from .mcp.json and delete the venv folder.
MCP
- Transport
- stdio
- Authentication
- not required
Security check
- It only reads local documentation and the index, with no writes to external systems
README in short
The README gives a table of the three servers with doc file counts, quick-start steps using jivo-docs-mcp as an example (venv, dependencies, index build, .mcp.json setup), the common layout of each server, a list of seven tools, and the tech stack (FastMCP, LlamaIndex, FAISS, sentence-transformers).
FAQ
Do I need to build the index myself?
Yes, faiss_index.bin and storage/ are excluded from the repository to keep its size down; builder.py creates them after cloning.
Does the repository cover other Sber platforms?
Currently only Jivo, SaluteBot and SberSalute Code; the README states documentation for other platforms will be added over time.
Related
Microsoft's utility and MCP server that turn PDF, Word, Excel, PowerPoint, HTML and other files into Markdown for models
Anthropic skills
Skills
Anthropic's official skill examples: docx, pdf, pptx and xlsx documents, design, MCP builder and the Agent Skills spec
A skill that builds editable PPTX decks from PDF, DOCX, URLs and Markdown with native shapes, charts and animations
A skill that draws editorial diagrams as standalone HTML and SVG files: architecture, flowcharts, ER, funnels and about forty more types