Repolode
Submit
nvk

nvk/llm-wiki

Claude Code, Codex and OpenCode plugin that builds LLM-compiled, Obsidian-compatible knowledge bases through parallel research, source ingestion and querying.

View on GitHubWebsite
Stars
1,435
Forks
135
Open issues
32
Last commit
18 days ago

What is it?

llm-wiki is a plugin and instruction set that lets AI coding agents maintain a personal or project knowledge base written in Markdown. Users capture ideas, run parallel research agents on a topic, ingest URLs, files, PDFs and bulk sources such as Git doc repos or MediaWiki dumps, and have the agent compile them into wiki articles that can be queried later.

The project ships packaging for Claude Code as a native plugin, for OpenAI Codex as a marketplace plugin, and for OpenCode, Pi and any other agent as a single instruction file. Shared behavior lives in one source protocol, and runtime-specific wrappers are generated from it. The wiki is stored as plain files and is compatible with Obsidian.

Beyond research and compilation, it provides thesis-driven investigation that weighs evidence for and against a claim, audits of generated outputs, reports, slides and study guides, dataset and inventory tracking, and an opt-in session memory layer with redacted capture.

Key features

  • 01Runs parallel research agents on a topic, with modes from quick to deep.
  • 02Ingests URLs, files, PDFs and bulk sources like Git doc repos and MediaWiki dumps.
  • 03Compiles raw sources into wiki articles and answers questions against them.
  • 04Installs as a native plugin for Claude Code and a marketplace plugin for Codex.
  • 05Stores the knowledge base as plain Markdown compatible with Obsidian.

Use cases

  • Building a long-running research wiki on a subject that grows over weeks.
  • Turning a folder of papers, articles and docs into a searchable, cross-referenced knowledge base.
  • Giving coding agents a persistent, queryable memory of a project's background material.

Quick start

  1. 1

    Install the Claude Code plugin

    claude plugin install wiki@llm-wiki
  2. 2

    Register the llm-wiki marketplace in OpenAI Codex

    codex plugin marketplace add nvk/llm-wiki
  3. 3

    Install the wiki plugin in Codex, then start a new thread and use @wiki

    codex plugin add wiki@llm-wiki

Things to know

  • Web search in OpenCode requires setting OPENCODE_ENABLE_EXA=1.
  • Running Codex inside a nono sandbox requires read-write access to $HOME/.codex.
  • If your wiki is not at the default ~/wiki path, you must configure the hub path.

Topics

#llm#knowledge-base#ai-agent#claude-code#codex#markdown#obsidian#research#agentic-ai#agentic-skills

FAQ

What is llm-wiki?
Claude Code, Codex and OpenCode plugin that builds LLM-compiled, Obsidian-compatible knowledge bases through parallel research, source ingestion and querying. llm-wiki is a plugin and instruction set that lets AI coding agents maintain a personal or project knowledge base written in Markdown. Users capture ideas, run parallel research agents on a topic, ingest URLs, files, PDFs and bulk sources such as Git doc repos or MediaWiki dumps, and have the agent compile them into wiki articles that can be queried later.
Is llm-wiki free to use? What license does it have?
llm-wiki is open source under the MIT license. Check the license terms before using it in commercial products.
Is llm-wiki still maintained?
Actively maintained. The last commit to llm-wiki was on Oct 2, 2026.
What can I use llm-wiki for?
Building a long-running research wiki on a subject that grows over weeks. Turning a folder of papers, articles and docs into a searchable, cross-referenced knowledge base. Giving coding agents a persistent, queryable memory of a project's background material.
How do I install llm-wiki?
Install the Claude Code plugin: claude plugin install wiki@llm-wiki Register the llm-wiki marketplace in OpenAI Codex: codex plugin marketplace add nvk/llm-wiki Install the wiki plugin in Codex, then start a new thread and use @wiki: codex plugin add wiki@llm-wiki
Where can I get llm-wiki?
The source code and installation instructions are on GitHub: https://github.com/nvk/llm-wiki

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

Related projects