The TUI

care (no subcommand) launches the Textual TUI. It opens directly into the chat surface — a Claude-Code-style transcript with a prompt at the bottom and a mode toggle above it.

The chat surface

• Type a task in natural language → MAGE generates a chain, CARL runs it, the answer prints inline.
• Type a slash command (/help, /library, …) for the non-chat affordances.
• The mode toggle above the prompt switches Interactive ↔ Production — Interactive is the default.
• Inspect the chain behind any answer: chat exposes chain DAG affordances to step through the graph, and an Artifacts pill collects the files a run produced.
• First-time users get a one-line offer to type /tour for a 5-step walkthrough.

How a run unfolds

As MAGE generates and CARL executes, the transcript shows a stage trail — ▶ Friendly Label… lines that complete to ✓, with ⎿ sub-rows for details. The status bar tracks token usage live.

Attaching files

Reference files inline with @<path> — e.g. summarise @report.pdf. PDFs are text-extracted; images are embedded for vision-capable models. Quoting works for paths with spaces: @"My Notes.md".

Global keys

Key	Action
Ctrl+P	Command palette (fuzzy over commands + saved entities).
Ctrl+B	Task list drawer (in-flight workers).
Ctrl+K	Capability catalog (skills / MCP / tools).
?	Help — tutorial + every binding for the active screen.
Ctrl+C	Quit.

Each screen layers its own bindings — press ? for the set filtered to the screen you’re on.

Where to go next

• Screens reference — every screen + modal and how to reach it.
• Slash commands — the full /command list.
• Chat modes — Interactive vs Production.
• Chain DAG — inspect the graph behind an answer.