What it does

This plugin wires Obsidian to a locally running Ollama instance. Highlight text (or use the full note), pick a command, and the model’s response drops in at your cursor. It ships with eight canned prompts—summarize, explain, expand, rewrite in a few styles, caption—and lets you roll your own with per-prompt model and temperature settings.

The interesting bit

The whole thing stays on your machine. No API keys, no token counting, no vendor lock-in. The README notes Ollama is “currently” Mac-only, which dates it slightly, but the plugin itself just hits localhost:11434—so if Ollama runs, it runs.

Key highlights

• Eight built-in prompts for common text operations
• Custom prompts with per-prompt model and temperature control
• Operates on selection or full note context
• Inserts output directly at cursor—no clipboard gymnastics
• Zero external API dependencies; local inference only

Caveats

• README still says Ollama is Mac-only, which hasn’t been true for a while; the plugin likely works fine on Linux/Windows Ollama installs, but this isn’t acknowledged
• No mention of how it handles long context windows or streaming responses

Verdict

Worth a look if you already live in Obsidian and want LLM assistance without the SaaS tax. Skip it if you need collaborative features, cloud sync, or don’t want to babysit a local model.