xknow-mcp
It is a local, private MCP server for searching, reading, exploring, and citing the XKnow SEO/SaaS/LLM-wiki knowledge base.
search_knowledge: ranked search over notes, returning snippets and URLs, with optional section filter and result limit.get_page: fetch the full text of a note by title or slug, preserving[[wikilinks]].explore_concept: retrieve a note plus outbound links and backlinks, optionally up to 2 hops.list_topics: list available notes grouped byseo,saas, orblog.cite: return the canonical citation (title, description, URL) for a note.Runs over stdio with a bundled free snapshot or a local purchased vault; no account, API key, or query-time network calls.
Note: README also lists
lint_rules, but the provided server schema does not expose it.
Click on "Deploy Server".
Wait a few minutes for the server to deploy. Once ready, it will show a "Started" state.
In the chat, type
@followed by the MCP server name and your instructions, e.g., "@xknow-mcp search XKnow for SaaS pricing models and summarize the trade-offs"
That's it! The server will respond to your query, and you can continue using it as needed.
Here is a step-by-step guide with screenshots.
xknow-mcp
XKnow · MCP server page · Source · npm · MCP Registry
An MCP server for the XKnow Knowledge Base — a curated, cross-linked body of SEO, SaaS, and LLM-wiki knowledge. Give any MCP-capable agent (Claude, Cursor, Cline, opencode, and others) the ability to search and cite it.
Local and private. Runs on your machine over stdio. No server, no account, no API key.
Static data. The free layer is a snapshot of the public XKnow guides, bundled with the package. Nothing is fetched at query time.
Graph-aware. Beyond keyword search,
explore_conceptwalks the cross-links between notes, so an agent can follow a topic instead of getting a flat list.
Quick start
Add it to your MCP client.
Claude Desktop (claude_desktop_config.json), Cursor, Cline, and most clients:
{
"mcpServers": {
"xknow": {
"command": "npx",
"args": ["-y", "xknow-mcp"]
}
}
}Claude Code
claude mcp add xknow -- npx -y xknow-mcpopencode (opencode.json)
{
"mcp": {
"xknow": { "type": "local", "command": ["npx", "-y", "xknow-mcp"], "enabled": true }
}
}Then ask your agent something like:
"Use the xknow tools to explain keyword difficulty and cite the source." "Search XKnow for SaaS pricing models and summarise the trade-offs."
Related MCP server: workspace-kb
Tools
Tool | What it does |
| Ranked search over the knowledge base. Returns titles, snippets, and URLs. |
| Full text of one note by title or slug (preserves |
| A note plus its outbound links and backlinks — walks the knowledge graph. |
| Lists every note grouped by section ( |
| The XKnow SEO/SaaS content self-check rubric — each rule with priority (P0–P3), reasoning, fix, and the knowledge-base notes that back it. |
| Returns the canonical citation (title, description, URL) for a note. |
Two knowledge layers
Layer | Content | How |
Free (default) | The 57 public XKnow guides and blog posts | Bundled snapshot — just |
Full vault | Your purchased XKnow Knowledge Base (500+ linked notes) |
|
The full-vault mode reads a local folder of Markdown notes, so your purchased copy never
leaves your machine. Set it once with the XKNOW_VAULT environment variable if you prefer:
export XKNOW_VAULT="/path/to/SEO-SaaS-Vault"
npx -y xknow-mcpHow it works
xknow-mcp is a Node.js stdio server. The free layer is a static JSON snapshot generated
from the public site and bundled in the package (data/knowledge.json); search is a small,
dependency-free ranking pass over the bundled notes. Vault mode parses Markdown notes from
a local folder on the fly. No network calls, no vector database, no telemetry.
Development
npm install
npm run build:data # refresh data/knowledge.json from https://xknow.org/mcp/knowledge.json
npm run build # tsc
npm run smoke # exercise every tool over a real stdio connection
# also test against a local vault
SMOKE_VAULT=/path/to/SEO-SaaS-Vault npm run smokeLinks
Product page: https://xknow.org/mcp
Knowledge base: https://xknow.org/vaults/seo
Registry: https://registry.modelcontextprotocol.io/servers/io.github.techreone/xknow-mcp
License
MIT — see LICENSE.
Available Tools
5 toolsciteGet a citation for a noteA
Return the canonical citation (title, description, URL) for a note, so an answer can attribute the XKnow source.
Inputs: slug (required).
explore_conceptExplore a concept and its graphA
Return a note together with the notes it links to (outbound) and the notes that link back to it (inbound). Use to walk the knowledge graph around a concept instead of a flat search.
Inputs: concept (required), depth.
get_pageGet a knowledge-base noteA
Fetch the full text of one note by title or slug. Use after search_knowledge to read the note before answering. Preserves [[wikilinks]].
Inputs: slug (required).
list_topicsList knowledge-base topicsA
List the available notes, grouped by section (seo, saas, blog). Use to discover what the knowledge base covers.
Inputs: section.
search_knowledgeSearch the XKnow knowledge baseA
Search curated, source-backed SEO, SaaS, and LLM-wiki knowledge. Returns ranked notes with a snippet and URL. Use this to ground answers about search-engine optimization, SaaS business models, or AI knowledge bases.
Inputs: query (required), limit, section.
Tool Schema Changelog
Recent tool additions, removals, and schema changes observed during successful MCP inspections.
Loading tool schema changelog…
TDQS
Scored across 5 tools
Each tool targets a distinct operation: search, read, explore graph, list, and cite. No overlap in purpose or output, making selection straightforward.
Mostly consistent verb_noun patterns (search_knowledge, get_page, explore_concept, list_topics), but 'cite' deviates as a bare verb without an object. Minor inconsistency.
Five tools perfectly cover the core workflows of a knowledge retrieval server: discover, search, read, explore connections, and cite. Well-scoped with no extraneous tools.
Covers all essential operations for a read-only knowledge base. Missing tools for editing or adding notes, but the server appears curated and source-backed, so writes may be intentionally out of scope.
Maintenance
Related MCP Connectors
Cloud or self-hosted knowledge for AI agents: hybrid search, reranking, GraphRAG, scoped MCP tools.
Shared, peer-validated knowledge archive for AI agents — search, contribute, and validate via MCP
Shared knowledge base for AI agents. Semantic search across agents, no setup required — just a URL.
A second brain for your AI agent. Shared memory for your team's AI agents, plus public and private knowledge bases — works with ChatGPT, Claude, Cursor and any MCP client.
Related MCP Servers
- FlicenseNot gradedqualityDmaintenanceEnables AI agents to interact with a persistent knowledge graph backend using MCP tools for reading, searching, and analyzing wiki pages with vector search and graph algorithms.5-
- AlicenseNot gradedqualityBmaintenanceEnables agents to semantically search and read curated local markdown knowledge bases through MCP, with optional CLI and dashboard tooling.MIT
- AlicenseNot gradedqualityBmaintenanceEnables AI agents to perform local-first web search, fetch, crawl, extract, cache, research, and autonomous information gathering through MCP, with no API keys or cloud dependencies.841 npmAGPL 3.0
- FlicenseNot gradedqualityAmaintenanceEnables MCP-compatible AI agents to search, ask, and retrieve section-level cited answers from a local hybrid index over scattered notes, documents, and chat logs.-