ya-wiki-mcp 0.3.1

MCP server for Yandex Wiki API — pages, grids, YFM syntax, prompt templates

ya-wiki-mcp

MCP server for Yandex Wiki API. Works with Claude Code, Claude Desktop, and any MCP client.

Built on ya-wiki-api Python client.

Quick Start

# Add to Claude Code
claude mcp add ya-wiki \
  -e YA_WIKI_TOKEN=your-token \
  -e YA_WIKI_ORG_ID=your-org-id \
  -- uvx ya-wiki-mcp

# Or run directly
uvx ya-wiki-mcp

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "ya-wiki": {
      "command": "uvx",
      "args": ["ya-wiki-mcp"],
      "env": {
        "YA_WIKI_TOKEN": "your-token",
        "YA_WIKI_ORG_ID": "your-org-id"
      }
    }
  }
}

Features

• 36 tools — full CRUD for pages, dynamic tables (grids), resources, and tree navigation
• Page tree cache — local tree cache with auto-sync on page create/update/delete/clone, plus full refresh from API
• YFM reference — built-in Yandex Flavored Markdown syntax guide
• Markdown to YFM converter — automatically converts standard Markdown to Wiki format
• Prompt templates — manage reusable page templates directly from chat

Configuration

Variable	Required	Description
YA_WIKI_TOKEN	Yes	OAuth token. Get one at https://oauth.yandex.ru/
YA_WIKI_ORG_ID	Yes	Organization ID
YA_WIKI_ORG_TYPE	No	cloud (default) or business

You can set these in a .env file or pass via -e flags.

Getting a Token

1. Go to https://oauth.yandex.ru/ and create an app
2. Select "For API access or debugging"
3. Add scopes: wiki:write (full access) or wiki:read (read-only)
4. Get your token: https://oauth.yandex.ru/authorize?response_type=token&client_id=<ClientID>

Tools

Pages

Tool	Description
get_page	Get a page by slug or ID
get_page_content	Get page text only (no metadata)
create_page	Create a page (wysiwyg, page, grid, template). Auto-updates tree cache.
update_page	Update title, content, or redirect. Auto-updates tree cache on title change.
delete_page	Delete a page (returns recovery token). Auto-removes from tree cache.
clone_page	Clone a page to a new location. Auto-adds clone to tree cache.
append_content	Append text to a page (top, bottom, section, or anchor)
get_descendants	Get all subpages (all levels) by page ID
get_descendants_by_slug	Get all subpages (all levels) by slug

Resources

Tool	Description
get_page_resources	Get attachments, grids, and SharePoint docs for a page
get_page_grids	List grids attached to a page

Dynamic Tables (Grids)

Tool	Description
create_grid	Create a table on a page
get_grid	Get a table with filtering and sorting
update_grid	Update title or sort order
delete_grid	Delete a table
add_rows	Add rows
delete_rows	Delete rows
add_columns	Add columns
delete_columns	Delete columns
update_cells	Update cell values
move_rows	Move rows
move_columns	Move columns
clone_grid	Clone a table to another page

Page Tree Cache

Tool	Description
get_tree	Get tree from cache (default) or fresh from API (use_cache=false)
refresh_tree_cache	Full rebuild of tree cache from API by root slug
clear_tree_cache	Clear tree cache
suggest_placement	Suggest where to place a new page based on description
add_tree_section	Manually add a section to the tree
remove_tree_section	Manually remove a section from the tree
set_tree	Replace entire tree from YAML

Utilities

Tool	Description
convert_md_to_yfm	Convert Markdown to YFM (tables, callouts, details, underline, highlight)

Prompt Manager

Tool	Description
prompts_list	List all saved templates
prompts_get	Get and render a template with arguments
prompts_add	Create a template from chat
prompts_add_from_file	Load a template from a file
prompts_remove	Delete a template

Page Tree Cache

The server maintains a local tree cache at ~/.cache/ya-wiki-mcp/tree.yaml that maps your wiki's section hierarchy. This cache is shared across all terminal sessions and is used by get_tree and suggest_placement to help with page navigation and placement.

Auto-sync: The cache updates automatically when you create, update, delete, or clone pages through the MCP tools.

Manual refresh: Use refresh_tree_cache("root-slug") to rebuild the cache from the API. This fetches all descendant pages and their titles.

# First time — initialize cache from your wiki root
refresh_tree_cache("your-root-slug")

# Quick check — read from cache (instant)
get_tree()

# Force fresh — bypass cache and fetch from API
get_tree(use_cache=false)

# Reset — clear and re-fetch
clear_tree_cache()
refresh_tree_cache("your-root-slug")

Markdown to YFM Converter

Yandex Wiki uses YFM (Yandex Flavored Markdown), which differs from standard Markdown. The converter handles:

Markdown	YFM
| H1 | H2 | tables	#| || |# wiki tables
> [!NOTE] callouts	{% note info %}
<details><summary>	{% cut "Title" %}
<u>text</u>	++text++
<mark>text</mark>	==text==
<sup>text</sup>	text^super^
<sub>text</sub>	text~sub~

License

MIT