Hello everyone 
, my plugin, Steward, is now listed on Obsidian community plugin!
Steward is yet another AI-powered assistant that helps with search, vault management, and automation.
Start with a slash
Type / at the beginning of any empty line in your editor, or open the chat interface. Type your request and press Enter.
Type / ? for available commands. Press Shift+Enter for a new line. Type m: to switch models. The chat adapts to your theme and works in both editor and reading view.
Search anything
Built-in very fast search engine. Search by tags, properties, keywords, or file names. Uses TF-IDF scoring with typo tolerance. Results are paginated and scored by relevance.
Manage your vault
Create, update, delete, copy, and move files through simple commands. These operations work on artifacts created from previous steps. If artifacts are created locally, they remain local during operations.
Update directly in the editor:
Update the selection:
Generate media
Generate images and audio directly in your vault.
Read only what you need
Read specific sections, not entire notes. If you query: “read the table above” then Steward reads only the table above your cursor, not the entire note.
Build your own commands
Create custom commands that combine multiple actions into reusable workflows.
In-note definitions: Write commands as YAML blocks in the Steward/Commands folder. Define what they do, which commands to run, and how to use them.
LLM-assisted creation: Share your idea in plain language, and Steward helps create the command structure.
Fine-turn your command with customizing system prompts and excluding tools.
Automated workflows: Commands trigger automatically when files are created in specific folders, when notes are tagged certain ways, or when content matches patterns. Turn your vault into a self-organizing system.
Create User-defined command with LLM’s help:
Automated command:
Other features
Model fallback: Automatically switches to backup models when one fails or times out.
Intent classification: Uses embeddings to learn from queries. Similar requests are cached for faster responses and fewer tokens.
Fast startup: Lazy-loaded services mean quick startup even with large vaults. Internal services initialize only when needed.
Get started
Install from Obsidian Community Plugins, configure API keys, and start typing / . Works with OpenAI, Gemini, DeepSeek, Ollama, and other popular models.
NOTE: This is the first version of Steward. Bugs are expected, and contributions are welcome!
Check my repo: Steward