Repolode
Submit
SamurAIGPT

SamurAIGPT/llm-wiki-agent

A coding agent skill that turns raw source documents into a persistent, interlinked markdown wiki maintained by Claude, Codex, OpenCode or Gemini CLI.

Open-source alternative to:Notion AI

View on GitHub
Stars
3,679
Forks
428
Open issues
13
Last commit
5 days ago

What is it?

LLM Wiki Agent is a personal knowledge base that you build by dropping source files into a raw/ folder and asking a coding agent to ingest them. The agent reads each source, writes a summary page, creates entity and concept pages for the people, companies, projects and ideas it mentions, and updates an index, a log and a running overview. Cross-references are added as [[wikilinks]], and contradictions with earlier sources are flagged at ingest time.

It runs inside Claude Code, Codex, OpenCode or Gemini CLI, each reading its own schema file (CLAUDE.md, AGENTS.md or GEMINI.md) that defines the page formats and workflows. No separate API key or Python setup is required for the agent path. Non-markdown files such as PDF, DOCX, PPTX and XLSX are converted to markdown through markitdown at ingest time. Output is plain files, so the wiki works as a git repository and can be opened in Obsidian.

The project also builds a knowledge graph: wikilinks become extracted edges, the agent adds inferred relationships, Louvain community detection groups related nodes, and the result is written to a self-contained graph.html that opens in a browser.

Key features

  • 01Turns dropped source files into structured summary, entity and concept pages.
  • 02Maintains an index, an append-only operation log and a running overview page.
  • 03Flags contradictions between a new source and existing claims at ingest time.
  • 04Converts PDF, DOCX, PPTX, XLSX, HTML and other formats to markdown automatically.
  • 05Builds an interactive graph.html from wikilinks and inferred relationships, with community clustering.
  • 06Works with Claude Code, Codex, OpenCode and Gemini CLI through their schema files.

Use cases

  • Following a research topic over weeks by filing papers and articles into one linked wiki.
  • Working through a book chapter by chapter, with characters and themes tracked as pages.
  • Keeping meeting notes, customer calls and project docs cross-referenced for a team.
  • Building a personal journal and reading-notes knowledge base that can be queried later.

Quick start

  1. 1

    Clone the repository

    git clone https://github.com/SamurAIGPT/llm-wiki-agent.git
  2. 2

    Enter the project directory

    cd llm-wiki-agent
  3. 3

    Open the project in your agent (e.g. Claude Code)

    claude

Things to know

  • Requires an existing coding agent such as Claude Code, Codex, Gemini CLI or OpenCode.
  • Non-markdown conversion relies on optional packages like markitdown, and some tools need a separate Python setup.

In collections

Open-source AI agent frameworks & appsStatic site generators & documentation tools

Topics

#knowledge-base#llm#markdown#personal-knowledge-management#knowledge-graph#obsidian#ai-agent#note-taking#ai-tools#automation

FAQ

What is llm-wiki-agent?
A coding agent skill that turns raw source documents into a persistent, interlinked markdown wiki maintained by Claude, Codex, OpenCode or Gemini CLI. LLM Wiki Agent is a personal knowledge base that you build by dropping source files into a raw/ folder and asking a coding agent to ingest them. The agent reads each source, writes a summary page, creates entity and concept pages for the people, companies, projects and ideas it mentions, and updates an index, a log and a running overview. Cross-references are added as [[wikilinks]], and contradictions with earlier sources are flagged at ingest time.
Is llm-wiki-agent free to use? What license does it have?
llm-wiki-agent is open source under the MIT license. Check the license terms before using it in commercial products.
Is llm-wiki-agent still maintained?
Actively maintained. The last commit to llm-wiki-agent was on Nov 2, 2026.
What can I use llm-wiki-agent for?
Following a research topic over weeks by filing papers and articles into one linked wiki. Working through a book chapter by chapter, with characters and themes tracked as pages. Keeping meeting notes, customer calls and project docs cross-referenced for a team. Building a personal journal and reading-notes knowledge base that can be queried later.
How do I install llm-wiki-agent?
Clone the repository: git clone https://github.com/SamurAIGPT/llm-wiki-agent.git Enter the project directory: cd llm-wiki-agent Open the project in your agent (e.g. Claude Code): claude
Where can I get llm-wiki-agent?
The source code and installation instructions are on GitHub: https://github.com/SamurAIGPT/llm-wiki-agent

Summary generated by AI from the project's README. GitHub stats refresh daily.

Related projects