Search...Search plugins and themes...
⌘K
Sign in
  • Get started
  • Download
  • Pricing
  • Enterprise
  • Account
  • Obsidian
  • Overview
  • Sync
  • Publish
  • Canvas
  • Mobile
  • Web Clipper
  • CLI
  • Learn
  • Help
  • Developers
  • Changelog
  • About
  • Roadmap
  • Blog
  • Resources
  • System status
  • License overview
  • Terms of service
  • Privacy policy
  • Security
  • Community
  • Plugins
  • Themes
  • Discord
  • Forum / 中文论坛
  • Merch store
  • Brand guidelines
Follow us
DiscordTwitterBlueskyThreadsMastodonYouTubeGitHub
© 2026 Obsidian

Intelligence Assistant

qwai-techqwai-tech56 downloads

AI chat, agents, and MCP tooling for your notes.

Add to Obsidian
  • Overview
  • Scorecard
  • Updates10

Transform your vault into an AI-native workspace. Intelligence Assistant brings a fast chat interface, configurable agents, MCP tool integrations, RAG over your notes, web search, CLI tools, and HTTP/OpenAPI tools — all without leaving Obsidian.

Desktop only. Requires Obsidian v1.7.2+.

✨ Feature Highlights

  • Modern Chat View — Streaming chat with provider/model badges, token usage display, tool-call visualization, and an Agent mode that runs multi-step agentic loops with native function calling.
  • Multiple LLM Providers — OpenAI, Anthropic, Google Gemini, DeepSeek, Ollama (local), OpenRouter, and SAP AI Core. Add your own credentials per provider.
  • Configurable Agents — Define reusable agents with custom prompts, tool permissions, model strategy (default/chat-view/fixed), RAG/Web/ReAct toggles, and MCP server access.
  • MCP Integration — Connect any Model Context Protocol server. Tool catalogs are cached for instant reuse. Includes a built-in MCP inspector for live testing.
  • RAG — Index your vault with a local vector store and inject relevant notes as context on every query. Uses real embedding APIs (OpenAI, Google, Ollama, DeepSeek, OpenRouter).
  • Web Search — Google CSE, Bing, Brave, SerpAPI, Tavily, SearXNG, Qwant, and Mojeek. Configurable locale, freshness, and domain filters.
  • HTTP / OpenAPI Tools — Point the plugin at any OpenAPI spec (local or remote); every path/verb pair becomes an agent tool automatically.
  • CLI Tools — Define custom shell commands as agent-callable tools with parameter templates, env vars, and platform presets.
  • Quick Actions — Pre-configured editor context menu actions (summarize, explain, fix grammar, improve writing, expand text).
  • Context Attachments — Attach files, images, or vault references to any message.

🚀 Quick Start

📦 Community Plugin (recommended)

  1. Open Settings → Community plugins in Obsidian.
  2. Browse for "Intelligence Assistant" and install.
  3. Enable the plugin, then open Settings → Intelligence Assistant to configure providers.

🔧 Manual Install

  1. Download the latest release.
  2. Extract into <vault>/.obsidian/plugins/intelligence-assistant/.
  3. Reload Obsidian and enable the plugin.

📋 Requirements

Requirement Details
Obsidian v1.7.2 or later (desktop only)
Node.js 18+ with npm (for building from source)
LLM API key Per provider (OpenAI, Anthropic, Google, DeepSeek, OpenRouter, etc.)

💬 Chat Experience

  1. Open the chat via the ribbon icon or Command Palette → Open AI chat in sidebar.
  2. Choose Chat or Agent mode in the toolbar. Agent mode runs the model in an agentic loop with native function calling and tool execution.
  3. Pick a provider, model, temperature, and token cap from the header controls.
  4. Toggle RAG (vault context) or Web search in the input bar.
  5. Attach files or vault references with @filename.
  6. Every response shows the provider, model, and token usage. Use Insert to Notes to drop a reply into the active file.

🤖 Agents

Defined under Settings → Agents. Each agent has:

  • System prompt, tool permissions, MCP server access
  • Model strategy: default (use settings default), chat-view (follow chat selector), or fixed (specific model)
  • Capabilities: RAG, Web Search, agentic tool-call loops with configurable max steps
  • Custom icon and display name

Agent mode uses native function calling (OpenAI) with automatic fallback to text-based tool parsing for other providers.

🔌 MCP Tools

Connect MCP servers under Settings → MCP. The plugin:

  • Connects at startup and caches the tool manifest
  • Injects MCP tools into agent loops automatically
  • Provides a live MCP Inspector for testing tools interactively
  • Supports stdio transport

🌐 Web Search

Configure under Settings → Web Search. Pick one of 8 supported providers (Google, Bing, Brave, SerpAPI, Tavily, SearXNG, Qwant, Mojeek) and set locale, result freshness, and domain filters. Credentials are stored per vault.

📡 HTTP / OpenAPI Tools

Add OpenAPI specs under Settings → Tools → HTTP / OpenAPI. For each source:

  • Point at a local .json file or a remote URL (cached locally)
  • Override the base server URL and inject auth headers or query params
  • Enable/disable the source; reload tool definitions on demand

⌨️ CLI Tools

Define custom shell commands under Settings → Tools → CLI Tools. Each tool supports parameter templates ({{param}}), argument/env insertion modes, working directory, timeout, and platform-specific presets (25+ built-in presets for file, search, network, code, data, and system operations).

🌐 Internationalization

The plugin supports all 46 Obsidian languages. Locale is detected automatically from your Obsidian language setting. Translations are community-driven — feel free to improve them via PR.

中文文档 | English

🛠️ Development

npm install          # install dependencies
npm run dev          # development build + file watcher
npm run lint         # ESLint (src/ + main.ts)
npm run type-check   # TypeScript type check (no emit)
npm run test         # Jest test suite
npm run build        # production bundle

Additional scripts: npm run dev:deploy (build + deploy to local vault), npm run deploy:local.

✅ Post-task rule

After every change, run npm run lint && npm run build. Both must succeed before the work is considered done.

📖 Documentation

Document Description
README-zh.md Chinese README (中文说明)
docs/README.md Documentation index
docs/architecture/overview-en.md Architecture overview (English)
docs/architecture/overview-zh.md Architecture overview (Chinese)
docs/project/project-guide-en.md Developer-oriented project guide (English)
docs/project/project-guide-zh.md Developer-oriented project guide (Chinese)
docs/reference/project-structure.md Full source tree reference

Contributions, issues, and feature requests are welcome — open a PR or discussion.

85%
HealthExcellent
ReviewSatisfactory
About
Transform your vault into an AI-native workspace with a streaming chat, provider/model badges, token usage display, and configurable agents across multiple LLM providers. Index notes for RAG, run CLI and MCP-backed agents, call web search and OpenAPI/HTTP tools, and attach files or vault references to messages.
AIAutomationSearch
Details
Current version
0.0.10
Last updated
2 days ago
Created
6 months ago
Updates
10 releases
Downloads
56
Compatible with
Obsidian 1.7.2+
Platforms
Desktop only
License
MIT
Report bugRequest featureReport plugin
Author
qwai-techqwai-tech
github.com/qwai-tech
GitHubqwai-tech
  1. Community
  2. Plugins
  3. AI
  4. Intelligence Assistant

Related plugins

Smart Connections

AI link discovery copilot. See related notes as you write. Lookup using semantic (vector) search across your vault. Zero-setup local model for embeddings, no API keys, private.

Semantic Notes Vault MCP

Give Claude Desktop and other AI assistants semantic access to your notes through a built-in Model Context Protocol (MCP) server.

Local LLM Helper

Use a local LLM server to augment your notes

QuickAdd

Quickly add new notes or content to your vault.

Templater

Create and use dynamic templates.

Homepage

Open a note, base, or workspace on startup, or set it for quick access later.

Omnisearch

Intelligent search for your notes, PDFs, and OCR for images.

Copilot

Your AI Copilot: Chat with Your Second Brain, Learn Faster, Work Smarter.

Claudian

Embeds Claude Code/Codex as an AI collaborator in your vault. Your vault becomes agent's working directory, giving it full agentic capabilities: file read/write, search, bash commands, and multi-step workflows.

Quick Switcher++

Enhanced Quick Switcher, search open panels, and symbols.