Reference documentation.
Docs.
Reference documentation for every feature. Start here, follow links to the detail you need.
“I want to chat with my notes.”
@-mention vault notes — `@`-attaching notes, Agent tools — vault-modifying tools
“I want to compare models.”
Multi-model arena — parallel streaming, Providers — the full provider list
“I want two chats at once.”
Multi-pane chat — splits and popout windows
“I want agents working for me in parallel.”
Swarm mode — leader mode and follower agents
“I want Curtis to run while I'm away.”
Scheduled runs — a prompt or agent on a daily/interval cadence, Desktop notifications — to get pinged when a background response lands
“I want Claude Desktop (or any AI app) working in my vault.”
MCP server — serving the vault over localhost MCP
“I'm writing an Obsidian plugin that needs AI.”
Plugin API — the public `api` object
“I want Curtis to read my cloud files.”
GCP connector — the GCP connector and its read-only Cloud Storage tools
“I want different AI workers with different jobs.”
Named agents — named agents — personas, model routing, tool ceilings
“I want to teach Curtis a reusable procedure.”
Skills — vault skills — write once, load on demand, portable to other agents
“I want a shell next to my notes — or agents that can run one.”
Terminal — the terminal pane and the command tool
“I'm bringing my history over from ChatGPT or Claude.”
Chat import & the `.curt` format — the importer and the `.curt` format
“I use Claude Code and want my setup here.”
Chat import & the `.curt` format — the one-click bring-over — MCP servers, skills, subagents, CLAUDE.md
“I want my chats as real files — synced, searchable, hand-editable.”
Chat import & the `.curt` format — the `AI/Conversations/` storage format
“I'm moving to a new machine or a second vault.”
Chat import & the `.curt` format — the full backup zip — chats, prompts, harnesses, settings (keys stripped) — or Chat import & the `.curt` format — chats only
“I want AI to edit my writing.”
Inline diff rewrite — diff-reviewed rewrites, Selection actions — direct-write transformations
“I want fully offline AI.”
Providers — Ollama / LM Studio setup
“I want voice input/output.”
“I want Curtis to remember things about me.”
“I want Curtis to know my notes without me pointing at them.”
Vault retrieval (RAG) — the hybrid retrieval index and auto-injected context
“Something isn't working.”
FAQ & troubleshooting — quick answers and fixes
“I'm developing the plugin.”