-
Notifications
You must be signed in to change notification settings - Fork 0
Home
A chat sidebar for Obsidian that works with any OpenAI-compatible API — DeepSeek, OpenRouter (300+ models including Claude, GPT and Gemini), or your own endpoint. Bring your own key. It reads and edits your notes, searches your vault, browses the web, and works with images — all from a conversation in the sidebar.
Real Chat is free and MIT-licensed. Source: saorgage/real-chat.
- Setup — install, pick a provider, paste your key.
- Usage — sessions, context, search, slash commands, inline edit, keys.
- Vision & Images — paste a picture and have the model see and act on it.
- Tools & Vault Actions — what it can do to your notes, plus web access.
- Memory & Triggers — persistent memory and custom commands.
- Five chat sessions, each with its own history, draft, model and settings. Histories persist across restarts; archive and reload past chats; export a chat to a note.
- Any OpenAI-compatible provider — DeepSeek direct (cache-aware pricing) or OpenRouter (live model list, per-model pricing, favourites). Per-session model selection.
- Acts in your vault — create, edit, append, rename, delete and find/replace notes; frontmatter and tag editing; backlinks, orphans and broken-link finders; word count and vault stats. File changes prompt for confirmation (or enable YOLO mode to skip).
- Ranked vault search — a 🔍 search box (toolbar button or hotkey-able command) for instant, token-free results; or ask in chat for synonym-expanded, reasoned search.
- Web access (optional) — web search, fetch a page, download a file into your vault.
- Images — paste, drag or attach; process locally (compress, resize, convert, crop, rotate); or have a vision model see them. See Vision & Images.
- Persistent memory & triggers, reasoning-model thinking blocks, cost tracking, streaming, slash commands, a saved prompt library, and inline "edit selection" from any note.
Real Chat only makes the network requests you configure or trigger. Your messages and any note content you add as context go to the API provider you choose; web access (off unless enabled) talks to your search provider and any URLs you/the model fetch. No telemetry, no other data collection. API keys live in the plugin's local data.json and are sent only to the provider they belong to. See Privacy & network use in the README.