WikiPeatia MCP Server
Server Configuration
Describes the environment variables required to run the server.
| Name | Required | Description | Default |
|---|---|---|---|
| WIKIPEATIA_TIMEOUT | No | HTTP timeout (seconds) for regular calls | 30 |
| WIKIPEATIA_DEEP_TIMEOUT | No | HTTP timeout for semantic / ask endpoints | 120 |
Instructions
Guidance the server publishes about itself, which clients place ahead of the tool catalog so the model reads it before choosing anything.
This server publishes no instructions, or was last inspected before Glama recorded them.
Capabilities
Features and capabilities supported by this server
Protocol revision2025-11-25
| Capability | Details |
|---|---|
| tools | {
"listChanged": true
} |
| logging | {} |
| prompts | {
"listChanged": false
} |
| resources | {
"subscribe": false,
"listChanged": false
} |
Tools
Functions exposed to the LLM to take actions
| Name | Description |
|---|---|
| wiki_searchA | Search wiki pages by title or content (MediaWiki search). Note: this is the wiki's built-in search; multi-word queries may return zero hits. For the full corpus (interviews, books, ...) use corpus_search. |
| wiki_get_pageC | Get the raw content of a wiki page. |
| wiki_get_page_summaryA | Get the first paragraph (intro extract) of a wiki page. Fast overview. |
| wiki_readC | Read a wiki page as plain text. With a section number, only that section. |
| wiki_suggestC | Autocomplete wiki page titles while typing. |
| wiki_list_categoriesC | List all categories on the wiki. |
| wiki_get_category_membersC | List all pages in a category. |
| wiki_list_all_pagesB | List all wiki pages alphabetically, optionally filtered by prefix. |
| wiki_list_libraryA | List all pages in the Library namespace ('Library:'). Ray Peat's books / newsletters / canonical texts live here. Read one with wiki_read('Library:01 Radiation And Growth', ...). |
| wiki_get_page_categoriesB | Show which categories a page belongs to. |
| wiki_get_page_linksB | List the wiki links contained in a page (related-topic discovery). |
| wiki_get_backlinksB | Which pages link to this page? (reverse link map) |
| wiki_get_page_sectionsB | List a page's section headings (table of contents). |
| wiki_get_page_infoC | Page metadata: last edit, length, last revision id, URLs. |
| wiki_get_randomB | Get random wiki page(s). Great for exploration. |
| wiki_get_recent_changesB | List recent edits on the wiki. |
| wiki_get_site_statsB | Wiki statistics: pages, edits, users, files. |
| corpus_healthA | Status and stats of the corpus search infrastructure. Returns: mode (hybrid), semantic_live (embedding search online), chunk_count, doc_count |
| corpus_searchA | Full-text search across the entire Ray Peat corpus. Covers articles, books (Library), interviews (audio transcripts), newsletters, forum posts, email Q&A and wiki pages. Results include excerpt, doc_id, doc_type, author, series, source_url, audio_url and timestamps (for interviews). |
| corpus_semantic_searchB | Semantic (embedding) search over the corpus. Finds passages close in meaning even without exact keyword matches; also returns glossary_terms (synonyms/concepts found for the query). |
| corpus_askA | Quick LLM-grounded answer over the corpus (fast, roughly 2-10 s). Returns a short synthesized answer with a confidence value and the cited chunks. For deep multi-source research with [S1][S2] citations, use corpus_deep_ask instead. |
| corpus_deep_askA | Deep, LLM-assisted question answering over the entire corpus. Synthesizes across all ~220K chunks; the answer carries [S1], [S2]... source markers. Open the underlying documents with corpus_get_doc(doc_id). Note: this endpoint is slow (30-120 s). Ask one question at a time. |
| corpus_suggestB | Corpus autocomplete: term and document (doc_id) suggestions. |
| corpus_get_docA | Fetch a document's full content (chunk by chunk). Interview chunks carry speaker and timestamp_start fields; use audio_url / source_url to reach the audio source. |
| corpus_list_doc_typesA | Document types and counts in the corpus (article, book, interview, ...). Usable as the doc_type filter in corpus_search. |
| corpus_list_authorsC | Authors in the corpus and their document counts. |
| corpus_list_seriesA | Series / collections in the corpus (e.g. 'One Radio Network', 'Ray Peat Newsletter', 'Ask the Herb Doctor'). Matches the 'series' field in corpus_search results; usable as a filter. |
| corpus_libraryA | Browse the corpus library (metadata list). For content use corpus_get_doc. |
| corpus_topicsA | List the corpus's topic tags (e.g. 'thyroid', 'pufa', 'cortisol'). Useful to see what themes the corpus is organized around. |
| raypeat_list_articlesA | List all of Ray Peat's articles on raypeat.com. Returns: Article list with titles and URLs. |
| raypeat_read_articleC | Read one of Ray Peat's articles as full text. |
| raypeat_search_articlesB | Keyword search across Ray Peat article titles. |
Prompts
Interactive templates invoked by user choice
| Name | Description |
|---|---|
No prompts | |
Resources
Contextual data attached and managed by the client
| Name | Description |
|---|---|
No resources | |
TDQS
Scored across 32 tools
Tools are mostly separable by prefix and resource, and descriptions explicitly distinguish similar ones (wiki_search vs corpus_search, corpus_ask vs corpus_deep_ask). However, there is still notable overlap among search/retrieval tools across wiki, corpus, and raypeat, which could cause occasional misselection.
All names use snake_case with consistent domain prefixes (wiki_, corpus_, raypeat_), and most follow prefix_verb_noun. Minor deviations like wiki_read and noun-only corpus_health/corpus_library are readable and do not break predictability.
32 tools is heavy for a single MCP server, even across three related domains; many retrieval/search variants could be consolidated or more clearly grouped. The count exceeds the recommended 15-tool range and increases selection burden.
The surface covers search, retrieval, metadata, discovery, QA, article listing, and corpus stats extensively, leaving few obvious retrieval gaps. It is read-oriented, so missing write/edit operations are acceptable for the apparent research purpose.