Yandex Wiki full-text search MCP server
Yandex Wiki Search MCP
An MCP server for Yandex Wiki: the same full-text search as the web UI, pages, comments, attachments and dynamic tables, 33 tools
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
- By default it can create, edit and delete wiki pages
- The OAuth token grants access to the organization's whole wiki within its permissions
Install
Manual install
uvx yandex-wiki-search-mcpLocal run; keys go into the WIKI_TOKEN and WIKI_ORG_ID variables.
This is third-party code. Review the repository files before installing.
What it does
The server gives 33 typed tools for Yandex Wiki: full-text search on the same backend as the web UI's search box (up to 50 results per query), a full page lifecycle (create, update, append at top, bottom or an anchor, clone, delete with a restore token), comments, file uploads, and 11 write tools for dynamic tables (grids): rows, columns, cells, copying and sorting. Every tool ships a JSON schema for input and output plus read-only, destructive or idempotent safety annotations. A server-side WIKI_READ_ONLY=true flag simply does not register the write tools, so the agent cannot call them at all. It runs over stdio for desktop clients and over streamable-http with Docker and optional multi-user OAuth for teams.
Who it is for. For teams keeping their docs in Yandex Wiki who want to search and edit pages through an agent.
Good fit when
- You need full-text search across the wiki with the same quality as the web UI
- You need to safely give an agent read-only access via WIKI_READ_ONLY without letting it edit pages
- You need to work with dynamic tables (grids): rows, columns, sorting
Not a fit when
- You need an official Yandex server: this is an unofficial project, not affiliated with or endorsed by Yandex
- You have no Wiki-scoped OAuth token and organization ID: the server will not connect without them
Example request
Find the onboarding process page in our Yandex Wiki and append a checklist section at the endLimitations
This is an unofficial project that runs on the same API as the web UI and is not endorsed by Yandex. Deleting a page returns a restore token, but write tools are active by default unless you explicitly set WIKI_READ_ONLY=true. The multi-user OAuth mode is only needed for a team HTTP deployment; local stdio use needs just one token.
How to disable. Remove the server from your MCP client config.
MCP
- Transport
- stdio, http
- Authentication
- OAuth
| Environment variables | |
|---|---|
| WIKI_TOKEN required, secret | Yandex OAuth token with Wiki access. |
| WIKI_ORG_ID required | Organization ID in Yandex 360. |
| WIKI_READ_ONLY | true disables registering write tools, making the server read-only. |
Security check
- By default it can create, edit and delete wiki pages
- The OAuth token grants access to the organization's whole wiki within its permissions
README in short
The Russian README with an English version describes 33 typed tools with JSON schemas and safety annotations, gives ready one-click install buttons for Cursor, VS Code, LM Studio and Claude Desktop with a preconfigured read-only mode, explains getting the OAuth token and organization ID, and covers Docker deployment with streamable-http and optional multi-user OAuth.
FAQ
How do I restrict the server to read-only?
Set the WIKI_READ_ONLY=true environment variable; write tools then are not registered and the agent cannot call them.
Is this an official Yandex server?
No, the README states plainly that the project is not affiliated with or endorsed by Yandex.
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