README file from
GithubSponsors
What's New
-
1.6- On-device local embedding models and multiple knowledge bases: index without any API key, split and manage knowledge bases independently, and let the Agent auto-pick the right one by name.
- CLI chat: on desktop, drive the Claude Code, Codex, Hermes, Pi, or Grok CLI you're already signed into from the same chat surface. Grok reuses an official Grok Build CLI login; it does not reuse an xAI API key.
- The new Learning Mode: turn any topic and reference material into a personalized learning project with structured outlines, knowledge points, flashcards, and an interactive knowledge map, backed by FSRS spaced repetition and Anki
.apkgimport for sustainable long-term review.
-
1.5: Introduces a new Agent runtime that turns AI from Q&A into active collaboration—with full tool calling, MCP, Skills, desktop Bash, subagents, and web search—plus smarter long-session context and memory, refreshed hybrid RAG, focus/PDF awareness, and multi-window chat with background Agents.
Highlights
Features
Beyond the core capabilities above, YOLO also provides:
| Feature | Description |
|---|---|
| 🖥️ CLI Agent (Desktop) | Reuse a supported locally signed-in CLI agent, including Claude Code, Codex, Hermes, Pi, and Grok, right inside Obsidian |
| 🔌 External Agent Support | Connect MCP clients such as Hermes and OpenClaw to YOLO's Vault search, or delegate tasks to a configured YOLO Agent |
| ⚡ Quick Ask | Ask, edit, and continue writing without leaving the editor |
| 🔎 Vault RAG | Retrieve across your Vault for answers grounded in your own notes |
| 🪟 Multi-Window Chat | Run different tasks and contexts in parallel across independent chat windows |
| 🧠 Memory System | Lets YOLO remember preferences, habits, and long-term context for more consistent conversations |
| 🪡 Cursor Chat | One-click context addition, conversation at your fingertips |
| ⌨️ Tab Completion | Real-time AI-powered completion for smoother, more natural writing |
| 🎛️ Multi-Model Support | OpenAI, Claude, Gemini, DeepSeek and other mainstream models, freely switch |
| 🌍 i18n | Native multi-language support |
Quick Start
- Open Obsidian Settings → Community Plugins → Browse → Search "YOLO"
- Install and enable
- Configure your API key in plugin settings, or use your own ChatGPT OAuth / Gemini OAuth:
- Open the sidebar to start chatting — or try Quick Ask by typing
@in the editor
For Grok subscription CLI chat on desktop, install Grok Build and run grok login (or grok login --device-auth) in a terminal first. YOLO asks the official CLI to reuse that cached login and does not copy its OAuth tokens into the Vault. The plugin starts a dedicated Grok ACP process in its default ask-first permission mode; YOLO auto-approval is not offered for this runtime.
Installation
Community Plugin Store (Recommended)
See Quick Start above.
Manual Installation
- Go to Releases and download the latest
main.js,manifest.json,styles.css - Create folder:
<vault>/.obsidian/plugins/obsidian-yolo/ - Copy files to that folder, then enable the plugin in Obsidian Settings
[!WARNING] YOLO cannot coexist with Smart Composer. Please disable or uninstall Smart Composer before using YOLO.
Roadmap
- Better and stronger Vault AI search
- Background Agent (long-running task automation)
- Multi-Agent orchestration (via subagents)
- Learning Mode — a dedicated study view
- Annotation Mode — real-time AI annotations and suggestions on notes
- Built-in assistant — a corner-pinned helper for config/agents, with auto-compaction and scheduled tasks
- Better AI whiteboard
- Voice input & meeting notes
Feedback & Issues
Hit a bug, something confusing, or have an idea? Open an issue:
🐛 Report a bug · ✨ Request a feature
What helps:
- Bug reports with a clear reproduction (Obsidian version, OS, plugin version, what you did, what happened)
- "I tried X and got Y" reports — UX papercuts, confusing wording, broken docs, outdated translations
- Concrete feature ideas tied to a real use case ("when I do A, I want B because C")
Please search existing issues first to avoid duplicates.
Contributing
All forms of contribution are welcome — bug reports, documentation improvements, feature enhancements.
Please open an issue first to discuss feasibility and implementation for major features.
See CONTRIBUTING.md for the full guide: what we welcome, AI-assisted PR policy, size guidelines, and dev setup.
Acknowledgments
Thanks to Smart Composer for the original work — without them, there would be no YOLO.
Special thanks to Kilo Code for their sponsorship. Kilo is an open-source AI coding assistant platform with 500+ AI models, helping developers build and iterate faster.
Support
If you find YOLO valuable, consider supporting the project:
Development logs are regularly updated on the blog.