Ar9av’s open-source obsidian-wiki project provides a framework for AI coding assistants to build and maintain an Obsidian-based “digital brain,” with documented support for Claude Code, Cursor, Codex and other tools, plus installable workflows for ingesting, querying, linting and updating notes.
Ar9av has published obsidian-wiki, an open-source framework for using AI coding assistants to build and maintain an Obsidian knowledge base.
The GitHub repository for Ar9av/obsidian-wiki describes the project as a framework for AI agents to create and maintain a “digital brain” through an Obsidian wiki. The repository says it is designed around Andrej Karpathy’s LLM Wiki pattern and can be used with Claude Code, Cursor, Codex and other AI coding assistants.
According to the GitHub project description, the system is meant to help agents record and update structured knowledge rather than leaving project context scattered across notes, code repositories and tool-specific memory files. Its stated goal is to give AI assistants a persistent, navigable reference base inside an Obsidian vault.
The GitHub repository says obsidian-wiki uses four ingest stages and delta tracking. In practice, that means the project is designed to process new or changed sources instead of repeatedly reprocessing everything in the vault.
The public materials do not verify some of the more specific claims circulating about automatic audits of saved tools, such as exact 12-hour recheck intervals or particular age-based rules for unused items. What the official repository does support is the broader idea: obsidian-wiki provides a structure for AI assistants to ingest, maintain and query an Obsidian-based knowledge base over time.
PyPI lists obsidian-wiki as a released Python package, with version 2026.7.8 shown on the project page. PyPI describes it as a package for building and maintaining an AI-maintained Obsidian knowledge base across AI coding agents, including Claude Code and Codex.
Skills.sh lists ar9av/obsidian-wiki with 32 skills. The catalog names installable workflows such as wiki-ingest, wiki-query, wiki-lint, daily-update and history-ingest, which indicates that the project is organized around reusable assistant actions rather than a single monolithic command.
TomeVault’s listing for the same repository says the configuration works with Claude Code, Codex, Copilot, Cursor, Gemini CLI and Windsurf. TomeVault also lists 32 skills and reports 33 of 33 audits passed for the configuration it displays.
Obsidian is already used by many developers and researchers as a local-first knowledge base. The notable part of obsidian-wiki is that it tries to make that vault usable by AI coding assistants as persistent project memory.
Instead of treating each assistant session as isolated, the framework’s documentation points toward a model where notes, history and project context can be maintained in a shared vault. That could be useful for teams or individuals who want AI tools to refer back to decisions, repository notes, prior work and documentation without relying only on chat history.
The available public sources support a narrower but still meaningful conclusion: obsidian-wiki is an open-source framework and packaged tool for connecting Obsidian with AI coding assistants, with documented workflows for ingestion, querying, linting and updates. Claims about specific private vault setups, exact schedules or custom audit policies should be treated as user implementations unless documented by the project itself.
Ar9av has published obsidian wiki, an open source framework for using AI coding assistants to build and maintain an Obsidian knowledge base.
What the project does The GitHub repository for Ar9av/obsidian wiki describes the project as a framework for AI agents to create and maintain a “digital brain” through an Obsidian wiki.
The repository says it is designed around Andrej Karpathy’s LLM Wiki pattern and can be used with Claude Code, Cursor, Codex and other AI coding assistants.
Continue reading
