AIAny
AI Client2026
Icon for item

claude-obsidian

Turns source documents into a linked, source-cited Obsidian wiki for grounded Q&A and research. Local-first vault workflows with content-addressed capture, provenance ledgers, BM25 retrieval, and Claude Code / Agent Skills integration.

Introduction

Most note-AI workflows lose the raw evidence or scramble provenance; that makes retrieval brittle and claims hard to audit. claude-obsidian treats the vault as first-class data: it captures immutable source copies, compiles them into linked Markdown pages, and surfaces provenance so every important claim can be traced back to original evidence. The result is a compounding, queryable knowledge base you own rather than a transient model context.

What Sets It Apart
  • Content-first, not model-first — it preserves content-addressed copies of sources before any synthesis, so summaries always point back to verifiable evidence (so what? you can audit and correct claims instead of trusting a transient LLM context).
  • Provenance and transaction safety — operations are planned, inspected, and applied as recoverable transactions with SHA-256 plans and journaling (so what? parallel workers produce drafts, but only one inspected bundle changes the vault, preventing silent overwrites and race conditions).
  • Vault-native outputs — generated notes are plain Obsidian Markdown, Canvas and Bases-friendly, and live in a normal filesystem vault (so what? your knowledge remains portable, searchable, and usable without a proprietary database).
  • Retrieval-first tooling — BM25 local retrieval, optional reranking, and explicit egress gates for model-based steps enable grounded Q&A workflows (so what? answers come from the vault evidence, reducing hallucination risk and making reviewable research reproducible).
Who it's for and tradeoffs

Great fit if you want an auditable, compounding personal knowledge base where an LLM does the filing rather than replacing your files. It's aimed at researchers, knowledge workers, and teams that need source-grounded summaries, evidence-led rollups, and explicit upgrade/migration controls.

Look elsewhere if you need a plug-and-play cloud sync note app or a simple transcript recorder — claude-obsidian deliberately avoids hidden caches and automatic uploads. It also assumes an Obsidian workflow, Python 3.11+ tooling, and occasional external runners for URL/YouTube/OCR tasks, so users seeking purely web UI or zero-setup hosted services may find it heavyweight.

Where it fits

This isn't a competitor to general-purpose hosted note apps; it's a local-first bridge between LLM-enabled workflows (Claude Code / Agent Skills) and durable, provenance-aware PKM. Think of it as the layer that makes RAG-style research auditable and maintainable inside an Obsidian vault.

Information

  • Websitegithub.com
  • AuthorsAgrici Daniel
  • Published date2026/04/07

Categories

More Items

GitHub
AI Client2025

Enhances web AI chat interfaces (Gemini, AI Studio, Claude, ChatGPT) with folder-based conversation organization, timeline navigation, prompt vault, cloud sync, chat export, UI fixes and a plugin engine. Runs as a browser extension and includes cross-site prompt management and Mermaid rendering.

GitHub
AI Agent2026

Manages real tmux-backed terminals and AI agents as draggable nodes on an infinite pan/zoom canvas, with a Trello-style kanban view, persistent sessions that survive restarts, mobile companion support, and a browser Server Edition for self-hosting.

GitHub
AI Agent2026

Runs locally to learn your tastes and proactively discover content across Bilibili, Xiaohongshu, Douyin, YouTube, X, Zhihu, Reddit and the open web. Local-first agent storing data in a local SQLite, with a browser extension, optional desktop backend bundling embeddings (bge-m3/Ollama), and conversational feedback to refine recommendations.