llm-wiki
A tool that compiles LLM-driven research into structured knowledge bases for any AI agent, capturing rough ideas, running parallel research, and promoting approved briefs into delivery projects.
llm-wiki builds LLM-compiled knowledge bases for an AI agent to use. It captures rough ideas, researches and shapes them, and lets a user explicitly promote an approved brief into a delivery project, alongside parallel research, collector catalogs, session memory, source ingestion, compilation, audits, querying, and artifact generation.
It ships as a plugin for Claude Code, OpenAI Codex, OpenCode, and other portable agents, and is compatible with Obsidian vaults. Recent releases add configurable lint exit gates, portable packaging across runtimes without relying on rsync, safer Windows session-state handling, project knowledge checkpoint exports for cross-topic handoffs, and a personal specialist framework of user-owned review methods with per-topic allowlists and provenance tracking.
It is written in Python and MIT licensed.