# llmwiki

> llmwiki is a free MCP server for knowledge and memory. Compile documents into a cited markdown wiki that agents can query, lint and review. It runs locally on your machine and is started by your AI client.

- **Kind:** MCP server
- **Pricing:** Free
- **Platforms:** CLI, self-hosted
- **Website:** https://llmwiki.atomicstrata.ai/guides/mcp-agent-integration
- **Repository:** https://github.com/atomicstrata/llm-wiki-compiler
- **MCP Registry name:** io.github.atomicstrata/llmwiki
- **Categories:** knowledge and memory, files and documents
- **Listing updated:** 2026-11-15

## About

llmwiki follows the LLM Wiki pattern: instead of rediscovering facts from raw files on every question, it compiles sources once into interlinked pages with citations, review state and retrieval metadata. Through MCP an agent can ingest a URL or file, run an incremental compile, ask questions, and check wiki health. Profiles such as AutoSci for research or Newsroom for editorial work add their own record types and review gates.

Some tools need no credentials, including `ingest_source`, `read_page`, `lint_wiki`, `wiki_status` and the fast `run_eval` suite. `compile_wiki`, `query_wiki` and `search_pages` need a provider key such as `ANTHROPIC_API_KEY` or `OPENAI_API_KEY`. `get_context_pack` builds a token-budgeted evidence pack and falls back to lexical ranking with a warning when embeddings are unavailable.

The server needs Node.js 24 or newer and runs with `llmwiki serve --root <project>`. It also exposes resources such as the index, concept pages and the latest eval report, plus tools to export and import Open Knowledge Format bundles. The latest npm release is 1.5.0.

## Features

- **Incremental compile:** Turns sources into interlinked, citation-traceable pages and recompiles only what changed.
- **Evidence packs:** Builds token-budgeted packs with citations, freshness and warnings.
- **Review before publishing:** Answers can be saved as pages only after citation validation and a profile gate.
- **Wiki linting and eval:** Runs quality checks and scores wiki quality, with a credential-free fast suite.
- **OKF import and export:** Moves wikis as Open Knowledge Format bundles within project-confined paths.
- **Workflow actions:** Runs declared actions that stage writes but cannot approve human gates.
- **MCP resources:** Exposes index, concept, query, source, state and eval resources.

## Good for

- Build a cited research wiki from papers and notes
- Give Claude Code a compiled knowledge base to query instead of raw files
- Check which wiki pages are stale or orphaned
- Stage review candidates before publishing generated answers

## Install

### Claude Code (Terminal)

```sh
claude mcp add llmwiki -- npx -y llm-wiki-compiler
```

### Claude Desktop (claude_desktop_config.json)

```json
{
  "mcpServers": {
    "llmwiki": {
      "command": "npx",
      "args": [
        "-y",
        "llm-wiki-compiler"
      ]
    }
  }
}
```

### Cursor (~/.cursor/mcp.json)

```json
{
  "mcpServers": {
    "llmwiki": {
      "command": "npx",
      "args": [
        "-y",
        "llm-wiki-compiler"
      ]
    }
  }
}
```

### VS Code (.vscode/mcp.json)

```json
{
  "servers": {
    "llmwiki": {
      "type": "stdio",
      "command": "npx",
      "args": [
        "-y",
        "llm-wiki-compiler"
      ]
    }
  }
}
```

### Windsurf (~/.codeium/windsurf/mcp_config.json)

```json
{
  "mcpServers": {
    "llmwiki": {
      "command": "npx",
      "args": [
        "-y",
        "llm-wiki-compiler"
      ]
    }
  }
}
```

### Other clients (your client's MCP settings (most use this shape))

```json
{
  "mcpServers": {
    "llmwiki": {
      "command": "npx",
      "args": [
        "-y",
        "llm-wiki-compiler"
      ]
    }
  }
}
```

## Tools

- `ingest_source`: Fetch a URL or local file into sources/.
- `compile_wiki`: Run the incremental compile pipeline.
- `query_wiki`: Answer a question, optionally saving it as a page after citation validation.
- `search_pages`: Return full content of pages relevant to a question.
- `read_page`: Read a single page by slug.
- `lint_wiki`: Run quality checks and return structured diagnostics.
- `wiki_status`: Report page and source counts, stale and orphaned pages.
- `get_context_pack`: Build a token-budgeted evidence pack.
- `run_eval`: Score wiki quality with a fast or full suite.
- `export_okf`: Export the wiki as an Open Knowledge Format bundle.
- `import_okf`: Preview or stage an Open Knowledge Format bundle as review candidates.
- `verify_artifact`: Verify a hash-pinned artifact reference and return metadata.
- `list_workflow_actions`: List available workflow actions.
- `describe_workflow_action`: Inspect an action and its input contract.
- `run_workflow_action`: Run a declared action, capped at staged writes.
- `workflow_run_status`: Inspect a workflow run.

## FAQ

### What does it need to run?

Node.js 24 or later and a wiki project directory; some tools also need an LLM provider key.

### Which tools work without credentials?

ingest_source, read_page, lint_wiki, wiki_status, export_okf, import_okf and the fast run_eval suite.

### How do I start the server?

Run llmwiki serve --root followed by the wiki project path, or add it as an npx command in the client config.

### Which clients are documented?

Claude Desktop, Cursor and Claude Code.

### Is llmwiki free?

Yes. llmwiki is free to use.

### How do I add llmwiki to Claude Code?

Run this command in your terminal: claude mcp add llmwiki -- npx -y llm-wiki-compiler

### What are the best alternatives to llmwiki?

The closest alternatives to llmwiki are Codebase Memory, XERJ, Claude Code Ultimate Guide, Nautilus Compass Memory and note1.

## Alternatives

- [Codebase Memory](https://ai-powered.com/mcp/codebase-memory): Codebase Memory is an open-source MCP server for knowledge and memory.
- [XERJ](https://ai-powered.com/mcp/xerj): XERJ is an open-source MCP server for knowledge and memory.
- [Claude Code Ultimate Guide](https://ai-powered.com/mcp/claude-code-ultimate-guide): Claude Code Ultimate Guide is an open-source MCP server for ai models and llm tooling.
- [Nautilus Compass Memory](https://ai-powered.com/mcp/nautilus-compass-memory): Nautilus Compass Memory is an open-source MCP server for knowledge and memory.
- [note1](https://ai-powered.com/mcp/note1): note1 is an MCP server for productivity.
- [ASTra](https://ai-powered.com/mcp/astra): ASTra is a freemium MCP server for files and documents.
- [CARTESIAN ECU](https://ai-powered.com/mcp/cartesian-ecu): CARTESIAN ECU is a paid MCP server for files and documents.
- [MedXpert 图书馆连接器（通用版）](https://ai-powered.com/mcp/medxpert): MedXpert 图书馆连接器（通用版） is an open-source MCP server for knowledge and memory.

---
Source: https://ai-powered.com/mcp/llmwiki
