Commands

Cassady installs two equivalent binaries: cass and cassady. This page uses cass, but the same options and subcommands apply to cassady.

Top-level forms

cass [OPTIONS]
cassady [OPTIONS]
cass check [OPTIONS]
cass setup [OPTIONS]
cass --resume [CHAT_ID]

Run cass --help, cass check --help, or cass setup --help for the help generated by the current binary.

Startup behavior

cass starts a new interactive chat in the current directory.
cass --cwd PATH starts from PATH instead.
cass --resume CHAT_ID loads a saved chat.
cass --resume lists chats for the current directory.
If setup is incomplete, cass offers to run the interactive setup wizard before starting a chat.

On exit, Cassady prints a command like:

Resume this chat with: cass --resume <id>

Global options

--resume [CHAT_ID]: resume a chat, or list chats for the current cwd when no id is provided.
--model MODEL: use MODEL for this session.
--base-url URL: override the active provider's OpenAI-compatible base URL for this session.
--api-key-env ENV: read the API key from environment variable ENV for this session.
--cwd PATH: use PATH as the launch cwd and workspace root.
--readonly: force read-only mode.
--workspace-edit: force workspace-edit mode.
--full-access: force full-access mode.
--help: show help.
--version: show version.

The three access-mode flags conflict with one another.

Subcommands

`cass check`

Validates Cassady configuration under ~/.cass:

JSON syntax and schema.
duplicate provider/model ids.
model/provider references.
active provider and model resolution.
active API key availability.

Missing API keys for inactive providers are warnings. A missing active API key is an error and cass check exits with a non-zero status.

`cass setup`

Runs the interactive setup wizard in a terminal. It configures OpenAI-compatible providers, API key environment-variable references, and first models. It updates config.json, providers.json, and models.json while preserving unrelated entries where possible.

In-chat commands

Type / to open command autocomplete.

/model <model>: switch the model for future turns. Autocomplete lists models from ~/.cass/models.json.
/new: create a new chat for the current directory.
/resume <chat>: resume a saved chat from the current directory. Autocomplete lists matching chats.
/status: show chat id, state, model, access mode, cwd, record count, and current status.

Local commands can be used only when the agent is idle.

Keys

Enter: accept an autocomplete item when a menu is open; otherwise send the current message.
Ctrl-J or Ctrl-Enter: insert a newline.
Up/Down: move through autocomplete items when a menu is open; otherwise scroll the transcript.
Mouse wheel: scroll transcript.
PageUp/PageDown: scroll transcript by larger steps.
Shift-Tab: cycle access mode while idle: read-only → workspace-edit → full-access.
Tab: cycle reasoning effort while idle. Models that require reasoning skip off; models without reasoning metadata start at off.
Ctrl-O: toggle compact/full tool output display.
Ctrl-Shift-R or Ctrl-R: toggle reasoning display.
y: approve a pending tool approval prompt.
n or Esc: deny a pending tool approval prompt.
Esc: request cancellation while a turn is running.
Ctrl-C: request cancellation while busy; press twice within 1.5 seconds to exit.

Output notes

Tool calls are shown compactly by default. Press Ctrl-O to expand full tool output. Provider-streamed reasoning is hidden by default unless show_reasoning is enabled in config or toggled at runtime.

3.7 KiB Raw Blame History