# Curtis AI > Curtis is an AI chat and records product with a paper-ledger ethic: every conversation kept as a portable .curt file you own. Chat across AI providers on your own keys, port chats from other tools in-browser, and keep append-only receipts for agent activity. ## Products - [Curtis Chat](https://curtisai.tech/chat): The agent layer for Obsidian — 53 AI providers on your own keys. Named agents that work your notes, every conversation kept as a portable .curt file. v1.0.0, source-available, free to use. - [Curtis Porter](https://curtisai.tech/porter): Convert ChatGPT, Claude, Grok, and Gemini exports into portable Curtis records — right in your browser, with zero network calls. - [Curtis Audit](https://curtisai.tech/audit): Append-only receipts for everything AI agents do in your notes — sequenced, checkable, and kept as markdown files you own. - [The .curt standard](https://curtisai.tech/curt): An open, single-file conversation format for AI chats — byte-robust, strictly validatable, zero dependencies. Spec and reference implementation, MIT. ## Documentation - [Docs](https://curtisai.tech/docs): Reference documentation for every Curtis feature, rendered from the plugin's docs. - [The console](https://curtisai.tech/docs/console): The web console's field guide — records, keys, exporting. - [Obsidian platform](https://curtisai.tech/docs/obsidian): The agent layer in Obsidian — install, what it does, into the reference docs. - [Porter docs](https://curtisai.tech/docs/porter): The migration tool — exports to .curt, zero network. - [Audit docs](https://curtisai.tech/docs/audit): Append-only receipts for agent activity, as markdown you own. - [Zed docs](https://curtisai.tech/docs/zed): The Curtis agent layer in the Zed editor — in development, targeting Q4 2026. - [Harness profiles](https://curtisai.tech/docs/harness): The harness is the base contract composed under every system prompt — before your system-prompt extension and before any agent persona. It defines how the model sees… - [Agent tools](https://curtisai.tech/docs/agent): AI tools that read, create, and edit your vault notes. This is the tool loop — the harness every chat runs on. For named workers (persona + model + permissions), see… - [GCP connector](https://curtisai.tech/docs/gcp): Read-only Cloud Storage on your own Google Cloud project, as agent tools. - [Named agents](https://curtisai.tech/docs/agents): Every agent gets its own lane; Curtis coordinates. A lane is a persona + a model + a tool ceiling — bind it to any chat, or hand it to a swarm leader as a specialist. - [Skills](https://curtisai.tech/docs/skills): Skills are reusable capability packages from the open Agent Skills standard — the same format used by Claude Code, Cursor, Copilot, Gemini CLI, OpenCode, and other… - [Swarm mode](https://curtisai.tech/docs/swarm): One leader chat, many follower agents. The leader delegates; you watch the vault work. - [Scheduled runs](https://curtisai.tech/docs/scheduled-runs): Put an agent on a cadence. A task fires headlessly, a note lands in your vault. - [MCP server](https://curtisai.tech/docs/mcp-server): Curtis is an MCP client — it calls tools from servers you run. Server mode is the reverse: Curtis serves your vault as MCP tools so external AI apps (Claude Desktop,… - [Plugin API](https://curtisai.tech/docs/plugin-api): Curtis exposes a small programmatic surface so other Obsidian plugins can use it as the vault's AI layer — one place configures the providers, keys, memory, and tools;… - [Multi-model arena](https://curtisai.tech/docs/arena): Send one prompt to 2–4 models in parallel. Pick a winner. - [Multi-pane chat](https://curtisai.tech/docs/multipane): Two chats at once. Each pane picks its own model. - [Terminal](https://curtisai.tech/docs/terminal): A shell in its own optional window. And, if you opt in, a tool the agent can run commands with. - [Chat import & the `.curt` format](https://curtisai.tech/docs/import): Move your chat history from other AI tools into Curtis and continue without missing a beat. New in v2.0. - [Link previews](https://curtisai.tech/docs/link-previews): External links grow a favicon; vault paths the assistant mentions become tappable. - [Inline diff rewrite](https://curtisai.tech/docs/diff-rewrite): Cursor-style AI rewrite with an Accept/Reject diff modal. - [Inline autocomplete](https://curtisai.tech/docs/autocomplete): Type in any note, pause, and a dimmed continuation appears after the cursor. Tab accepts. - [Voice I/O](https://curtisai.tech/docs/voice): Talk to Curtis. Listen to Curtis. Hands-free. - [Desktop notifications](https://curtisai.tech/docs/notifications): Curtis pings you when a response lands — even after you've moved on. - [@-mention vault notes](https://curtisai.tech/docs/mentions): Type @ in chat → fuzzy-search your vault → attach note content as context. - [Cross-conversation search](https://curtisai.tech/docs/search): Find any message across every conversation, past and present. - [Image attachments](https://curtisai.tech/docs/images): Curtis.ai supports vision — drag a screenshot in, paste a photo, or pick from a file. The model reads it and responds. - [Memory](https://curtisai.tech/docs/memory): Curtis.ai remembers durable facts about you across conversations. This doc explains what gets stored, where, how it's used, how to manage it, and the editing UI new in… - [Vault retrieval (RAG)](https://curtisai.tech/docs/vault-retrieval): Curtis embeds your notes once, then quietly hands the model the relevant excerpts with every message — and semanticsearch becomes an agent tool. - [Selection actions](https://curtisai.tech/docs/selection-actions): Right-click any text selection in a note for AI-powered transformations. Each action runs the selection through your active model — replace-mode actions open a diff… - [Slash commands](https://curtisai.tech/docs/slash-commands): Type / in the chat input to open the autocomplete menu. Arrow keys to navigate, Enter/Tab to pick, Esc to dismiss. Or run /help to open a reference modal inside Obsidian. - [FAQ & troubleshooting](https://curtisai.tech/docs/faq): Quick answers to the questions that actually come up. Every answer links to the full story. - [Providers](https://curtisai.tech/docs/providers): Curtis.ai ships with 53 built-in providers and supports any OpenAI-compatible endpoint as a custom provider. This doc covers what's built-in, how each one authenticates,… - [Changelog](https://curtisai.tech/changelog): Release history for Curtis products. - [Pricing](https://curtisai.tech/pricing): Curtis AI pricing. ## The .curt standard - [.curt format](https://curtisai.tech/curt): One file, one conversation, full fidelity. Open standard, MIT. ## Company - [Contact](https://curtisai.tech/contact): Contact Curtis AI. - [Licenses](https://curtisai.tech/licenses): Licensing for Curtis products and the .curt standard. - [Privacy](https://curtisai.tech/privacy): Curtis AI privacy policy — local-first, BYOK, no sale of data. - [Terms](https://curtisai.tech/terms): Terms of sale and service for the hosted tier. - [Open books](https://curtisai.tech/books): Curtis, in real numbers.