Obsidian Notes Now Run Your Existing Claude Code and Codex Subscriptions Directly
Anyone already paying for Claude Code or Codex can point that same agent at their Obsidian vault without a second API key or subscription. Diff previews and native backlink handling remove the usual fear of letting an agent touch local notes.
Oh My Claudian, an MIT-licensed plug-in for Obsidian, reuses the coding agents you already have configured in a terminal. Instead of wiring a new API key into your notes app, it loads the same Claude Code, Codex, Cursor Agent, DeepSeek Harness, OpenCode, Grok Build, or Oh My Pi CLIs, using each tool's existing authentication, model catalog, permissions, slash commands, Skills, and MCP services. A Readiness check verifies the CLI environment and can even install or update supported CLIs from the settings screen.
Inside a sidebar chat, typing '@' lets you attach multiple notes, and long tool-call sequences collapse into expandable summaries. When a provider supports it, file edits show as a diff card. Selected text can be rewritten through an inline edit command that previews word-level changes before applying. Any useful part of a chat reply can be saved directly to a new or existing note, converted back to standard Markdown with one-click undo.
For Obsidian-specific data, the plug-in integrates native operations: agents can inspect backlinks, add or modify Frontmatter properties, rename or move files while updating library links, and send files to the trash. Collapsing the sidebar turns the chat into a floating Zen Mode panel above the note. Setup requires Obsidian desktop 1.7.2 or newer and at least one logged-in CLI; two common pitfalls involve PATH differences from a GUI launch and a leftover ANTHROPIC_API_KEY overriding Claude Code subscription login.
Because it borrows the CLI's existing credentials and model routing, this plug-in sidesteps the extra-subscription model most note-AI integrations push, which matters for anyone who already pays for a coding agent.
Showing word-level diffs before applying an inline edit addresses the specific risk that AI will silently corrupt notes, a barrier that often stops writers from letting agents edit prose.