CwikiMCP
# Apache Incubator CWiki MCP
MCP server for the Apache Incubator Confluence space:
<https://cwiki.apache.org/confluence/display/INCUBATOR>
The server is read-only and works against public wiki pages.
GET responses are cached locally by default for 30 days. Use `force_refresh=true` on read tools to bypass the cache for a single call.
## Install
```sh
python3 -m venv .venv
. .venv/bin/activate
pip install -e .
```
## Run
```sh
incubator-cwiki-mcp
```
For local development:
```sh
python -m incubator_cwiki_mcp.server
```
## Test
```sh
pip install -e ".[dev]"
make check
```
## Configure With Codex or Claude Desktop
Use the Python module entrypoint:
```json
{
"mcpServers": {
"incubator-cwiki": {
"command": "<Path to CwikiMCP>/.venv/bin/python",
"args": ["-m", "incubator_cwiki_mcp.server"],
"env": {
"CWIKI_BASE_URL": "https://cwiki.apache.org/confluence",
"CWIKI_SPACE_KEY": "INCUBATOR"
}
}
}
}
```
## Tools
- `cwiki_space_info`: return metadata for the configured Confluence space.
- `cwiki_list_pages`: list pages in the Incubator space.
- `cwiki_search_pages`: search pages with Confluence CQL. Use `title_only=true` to match page titles only (faster and more precise for finding a specific page by name).
- `cwiki_get_page`: fetch a page by title or `page_id`.
- `cwiki_get_children`: list child pages for a `page_id`.
- `cwiki_cache_info`: show local cache settings and size.
- `cwiki_clear_cache`: clear the local cache.
## Environment
- `CWIKI_BASE_URL`: defaults to `https://cwiki.apache.org/confluence`.
- `CWIKI_SPACE_KEY`: defaults to `INCUBATOR`.
- `CWIKI_CACHE_DIR`: defaults to `.cache/cwiki`.
- `CWIKI_CACHE_TTL_SECONDS`: defaults to `2592000` seconds (30 days). Set to `0` to disable caching.
TDQS
Scored across 7 tools
Each tool has a clearly distinct purpose: cache management (info/clear), page retrieval (by ID/title), children listing, page listing, page search, and space info. No overlap in functionality.
All tools follow a consistent 'cwiki_verb_noun' pattern (e.g., cwiki_list_pages, cwiki_clear_cache). The naming convention is uniform and predictable.
Seven tools is a reasonable number for a focused Confluence wiki server. The set covers essential read operations and cache management, though it lacks create/update/delete operations, which might be justified if the server is read-only.
The tool set covers page retrieval, listing, search, and space metadata, but notably missing are create, update, and delete operations for pages. This is a significant gap for a full wiki management surface, though the server may be intentionally read-only.