Obsidian plugin that lets LLMs actually touch your files

What it does

YOLO is an Obsidian plugin that embeds a full AI assistant into your note-taking workflow. It supports sidebar chat, inline quick-asks via @, tab completion, and a RAG-powered knowledge base that answers from your actual notes rather than generic training data. The agent can read your current file and cursor position, search the web, and execute multi-step tasks through configurable toolchains.

The interesting bit

The “Agent mode” is the real departure from typical chat plugins: LLMs get tool-calling capabilities and MCP integration to directly manipulate vault files, while “Skills” let you encapsulate SOPs as reusable one-sentence invocations. The plugin also tracks long-term memory across conversations—preferences, habits, context—so the assistant doesn’t reset to a blank slate each time.

Key highlights

• Multi-window chat for parallel task contexts
• Hybrid search indexing across the entire vault with upgraded relevance in v1.5.5
• Context pruning and compression for long-running conversations
• Deep PDF support with vision reading and region screenshots (v1.5.7)
• Multi-model support: OpenAI, Claude, Gemini, DeepSeek, Groq
• Native i18n with English, Chinese, and Italian translations

Caveats

• Cannot coexist with the Smart Composer plugin; you must disable or uninstall it first
• Mobile support is explicitly limited: the maintainer guarantees “usable” but not feature parity with desktop, and some workflows may behave inconsistently
• Background agents and cron scheduling are on the roadmap but not yet implemented

Verdict

Worth a look if you want AI that actually operates on your notes rather than just commenting from the sidelines. Skip it if you need reliable mobile feature parity or if you’re already committed to Smart Composer.