English | Chinese (Simplified) | Chinese (Traditional)

geminese brings both Gemini and Ollama into Obsidian. Use Gemini for native cloud workflows, or switch to a local Ollama model for private, vault-native planning and agentic work.

Local models matter here because they are not just chat backends. In geminese, an Ollama model can work with your vault as an agent: inspect files, search notes, draft edits, and help you plan changes before anything is written.

Gemini remains the full native path for users who want Gemini CLI integration, cloud models, and Gemini-specific features.

• Dual Model Families: Switch between Gemini cloud models and locally running Ollama models directly in the chat toolbar.
• Full Agentic Capabilities: Work with your vault as a real agent — read, write, edit, and search files without leaving Obsidian.
• Local-First Workflows: Use Ollama for private, vault-native planning and agentic work through the local HTTP runtime.
• No API Key Required for Gemini: Gemini CLI authenticates with your Google account and works with the free tier (60 req/min, 1000 req/day).
• Context-Aware: Automatically attach the focused note, mention files with @, exclude notes by tag, include editor selection, and access external directories for additional context.
• Vision Support: Analyze images by sending them via drag-and-drop, paste, or file path.
• Inline Edit: Edit selected text or insert content at cursor position directly in notes with word-level diff preview.
• Instruction Mode (#): Add refined custom instructions to your system prompt directly from the chat input when using Gemini.
• Slash Commands: Create reusable prompt templates triggered by /command, with argument placeholders and @file references.
• MCP Support: Connect external tools and data sources via Model Context Protocol servers (stdio, SSE, HTTP) when using Gemini-native workflows.
• Model Selection: Choose Gemini Auto, Pro, Flash, Flash Lite, or any installed Ollama model discovered from your local runtime.
• Plan Mode: Toggle plan mode via Shift+Tab — Gemini or Ollama can explore and design before implementing.
• Security: Permission modes — Agent (execute tools and edit files) and Plan (read-only planning), plus command blocklist and vault-scoped access.
• 10 Languages: English, Chinese (Simplified/Traditional), Japanese, Korean, Spanish, German, French, Portuguese, Russian.

Plan mode makes local workflows safer: review first, decide second, edit only when you switch to Agent mode.

• Obsidian v1.4.5+
• Desktop only (macOS, Linux, Windows)
• For Gemini models: Gemini CLI installed and a Google account (free tier works)
• For Ollama models: a local Ollama runtime with at least one installed model

macOS & Linux

npm install -g @google/gemini-cli

Windows

1. Install Node.js from nodejs.org. Make sure "Add to PATH" is checked during installation.
2. Open Command Prompt or PowerShell and install the CLI:

   npm install -g @google/gemini-cli

3. IMPORTANT: Fully restart Obsidian after installation to ensure it picks up the new environment variables.

Then authenticate (on Terminal):

gemini

Follow the prompts to sign in with your Google account.

Install and start Ollama on your machine, make sure the local HTTP API is available, and keep at least one model installed. If needed, you can change the Ollama endpoint later in Settings via Ollama base URL.

1. Download main.js, manifest.json, and styles.css from the latest release
2. Create a folder called geminese in your vault's plugins folder:

   /path/to/vault/.obsidian/plugins/geminese/
   

3. Copy the downloaded files into the folder
4. Enable the plugin in Obsidian Settings → Community plugins

Supercharge geminese with Obsidian Skills — agent skills that teach Gemini and local vault workflows how to work with Obsidian Markdown, Bases, JSON Canvas, CLI, and more.

Just open gemini in Terminal and say:

Help me install the Obsidian Skills plugin from https://github.com/kepano/obsidian-skills

Gemini will clone the repo and set everything up for you.

_{If AI can do it, why should a human bother? Leave this kind of thing to the AI.}

• If you find this plugin helpful, please consider giving it a ⭐ on GitHub — it really helps!

Two modes:

1. Click the bot icon in ribbon or use command palette to open chat
2. Select text + hotkey for inline edit

Use Gemini or Ollama to read, write, edit, and search files in your vault.

Check you're connected: If you get a reply in the chat, you're connected. You can ask e.g. “What model are you?” to confirm. The model selector shows Gemini options and any available Ollama models; click it to switch. Permission mode (Plan / Agent): Plan is read-only planning, Agent allows tool execution and file editing.

• File: Auto-attaches focused note; type @ to attach other files
• Selection: Select text in editor, then chat — selection included automatically
• Images: Drag-drop, paste, or type path
• External contexts: Click folder icon in toolbar for access to directories outside vault

• Inline Edit: Select text + hotkey to edit directly in notes
• Instruction Mode: Type # to add refined instructions to system prompt
• Slash Commands: Type / for custom prompt templates
• MCP: Add external tools via Settings → MCP Servers; use @mcp-server in chat to activate

Customization

• User name: Your name for personalized greetings
• Excluded tags: Tags that prevent notes from auto-loading
• Media folder: Configure where vault stores attachments for embedded image support
• Custom system prompt: Additional instructions appended to the default system prompt

Safety

• Enable command blocklist: Block dangerous bash commands (default: on)
• Blocked commands: Patterns to block (supports regex, platform-specific)
• Allowed export paths: Paths outside the vault where files can be exported

Environment

• Custom variables: Environment variables (KEY=VALUE format)
• Environment snippets: Save and restore environment variable configurations

Advanced

• Gemini CLI path: Custom path to Gemini CLI (leave empty for auto-detection)

• Agent mode: Default mode — execute tools and edit files (with safety interception and approval)
• Plan mode: Read-only — explores and designs a plan before implementing

• Sent to API: Your input, attached files, images, and tool call outputs go to Google's Gemini API via the CLI.
• Local storage: Settings and session metadata stored in vault/.gemini/; session data managed by Gemini CLI.
• No telemetry: No tracking beyond Google's Gemini API.

If you encounter Gemini CLI not found, the plugin can't auto-detect your installation.

Solution: Find your CLI path and set it in Settings → Advanced → Gemini CLI path.

Alternative: Add your Node.js bin directory to PATH in Settings → Environment → Custom variables.

Make sure you've authenticated with Gemini CLI first:

gemini

This will open a browser for Google account login. After signing in, the CLI (and plugin) can use your account.

npm run dev     # Watch mode
npm run build   # Production build
npm run test    # Run tests
npm run lint    # Lint code

Obsidian Plugin (UI)
      ↓
child_process.spawn("gemini", ["--output-format", "stream-json", ...])
      ↓
Gemini CLI → Google Account (no API key)

The plugin spawns the Gemini CLI as a subprocess for each query, passing --output-format stream-json to get structured JSONL output. Session continuity is maintained via --resume.

See ARCHITECTURE.md for detailed source structure and development notes.

Based on Claudian — converted from Claude Code CLI to Gemini CLI. Uses your Google account (no API key needed).

Licensed under the MIT License.