bahgs/self-improving-obsidian-llm-wiki

Self-Improving Obsidian LLM Wiki is a template for building a personal knowledge base in Obsidian, a note-taking app that stores everything as plain Markdown files, where an AI agent does the work of reading source material you drop in and organizing it into linked notes, concept summaries, and structured wiki pages. The key idea is that the system improves itself over time. Most AI note systems just get bigger as you add material. This one is designed to also get better at its own process: as it ingests sources, it can propose improvements to its own operating rules, extraction patterns, and quality rubrics stored in a "schema" folder. Future ingest sessions run according to updated rules, so the system's judgment and output quality can evolve. The folder structure separates three layers: a "raw" folder for original source documents (which are never edited by the agent), a "wiki" folder for generated notes, concept pages, entity entries, and summaries, and a "schema" folder that contains the agent's operating instructions and its improvement log. You point any AI coding assistant (the template includes configuration files for Claude Code, Gemini, Cursor, and a generic agent format) at the repository root, tell it to read the agent instructions, then give commands like "INGEST: raw/web/example.md" or "QUERY: What are the strongest concepts in the vault?" You would use this if you accumulate a lot of research material, articles, papers, notes, book excerpts, and want an AI to maintain a connected knowledge graph from it without you manually organizing each piece. The full README is longer than what was provided.