Interactive Commands

All commands are typed at the TUI prompt, prefixed with /. Unknown slash commands are submitted to the agent as regular text (useful for paths starting with /).

Session Management

CommandDescription
/newStart a new session (autosaves current if non-empty)
/continueResume the most recent session (like cast -c, but mid-session)
/sessionsSession picker with type-to-filter search (by message text, project path, or id); switch or delete
/clearClear conversation context (and save the cleared state)
/compactForce context compaction now (auto-triggers near the limit)
/copyCopy last assistant response to clipboard
/currentShow all status bar data (even disabled segments)
/quit, /exitSave and exit

Model and Provider

CommandDescription
/modelOpen model picker (shows current model)
/model <name>Switch to a specific model (validated)
/subagent-modelOpen model picker for sub-agents
/subagent-model <name>Switch sub-agent model
/reasoningChange reasoning level (opens picker if model supports it)
/providerOpen provider picker (switch, add, or delete providers)
/provider addAdd a new provider (name → URL → key wizard)
/provider deleteDelete a provider
/provider <name>Switch to a named provider

Persona

CommandDescription
/personaOpen persona picker
/persona <name>Switch to a specific persona

See Personas for the full list.

Skills and MCP

CommandDescription
/skillsToggle skills on/off (multi-select). Also: list, enable/disable, uninstall, help
/skill:<name> [args]Force-load and run a skill by name
/pluginToggle installed plugins. Palette also has install / list / enable / uninstall / marketplace / help
/mcpToggle MCP servers on/off. Also: list, enable/disable, uninstall, help
/reloadRe-scan skills, rules, MCP servers, and personas for cwd

Bare /skills / /mcp / /plugin = multi-select toggle. list is read-only. uninstall always confirms (picker or typed). See Skills, MCP, Plugins.

Hot-reload vs /reload

You never need to quit cast or start a new session for these changes. The current chat continues.

ChangeApply how
/skills / /mcp / /plugin toggle, enable / disableAutomatic (hot-reload)
/plugin install / uninstall, /skills uninstall, /mcp uninstallAutomatic
/plugin marketplace remove (drops installed packs)Automatic (skills reload)
New/edited files on disk: skills, mcp.json, rules, personas, context files (including npx skills add)/reload in the same session

/reload only re-scans cwd resources — it does not reset the conversation.

Rules

CommandDescription
/rulesList loaded rules with their apply mode, globs, and scope
/rule:<name>Invoke a rule by name (loads its full content into context)

See Rules for rule types and creation.

Plan Mode

CommandDescription
/planEnter plan mode (explore and plan only, no code changes)
/buildExit plan mode, restore full toolset
/plan-model [name|off]Show/change the model used while plan mode is active

See Plan Mode for the full workflow.

Steering

These commands work while the agent is running:

CommandShortDescription
/steer <message>/sInject a message into the running turn
/queue <message>/qQueue a message for after the current turn
/queue-reset/qrClear the message queue
/abort, /stopStop current agent run

/steer interrupts the current turn with new context — the message is injected immediately into the conversation, and the agent sees it on the next tool-call iteration. Useful for correcting course mid-execution.

/queue saves a message to run after the agent finishes its current turn. The message becomes a new turn automatically.

If nothing is running, both /steer and /queue submit the message as a normal prompt.

/abort stops the current run and clears both the steering and follow-up queues — anything queued before the abort is discarded.

Both steering and follow-up messages reset the doom loop counter — repeating a failing command after user guidance is treated as a new attempt, not a loop.

Context and Usage

Token usage and context size are shown automatically in the TUI status bar (prompt tokens in, completion tokens out, context percentage, tokens/second, and sub-agent tokens).

Use /statusbar to toggle individual segments on/off and reorder them — useful on narrow terminals where the full bar overflows. Segments can be moved between the left and right sides of the bar with ←/→, and reordered within each side with j/k. Default: persona, mode, model (left) and elapsed (right); enable others via /statusbar.

CommandDescription
/usageShow cumulative session token/cost usage (prompt, completion, cache hits, sub-agent tokens)

Configuration

CommandDescription
/permissionsOpen permission mode picker
/permissions defaultSwitch to gated mode (confirm dangerous commands)
/permissions bypassSwitch to bypass mode (no confirmation)
/webToggle web tools (web_search, web_fetch) on/off
/web-search-providerSwitch the web_search backend between DuckDuckGo (free, rate-limited), Tavily (API key, 1000 free/month), and Brave Search (API key)
/statusbarToggle and reorder status bar segments (multi-select picker)
/themeOpen theme picker
/theme <id>Switch to a specific theme

Utility

CommandDescription
/repoShow cwd, git branch, dirty state, remote, and HEAD
/keysList all keybindings
/helpShow the command list
/sshManage SSH hosts — list, add, remove (persists to ~/.cast/ssh.json)

Keybindings

ActionKeys
Cursor up/down/left/right↑ / ↓ / ← / →
Word left/rightAlt+← / Alt+→ (or Ctrl+← / Ctrl+→)
Line start/endHome / End (or Ctrl+A / Ctrl+E)
Delete charBackspace / Delete
Delete wordCtrl+W / Alt+Backspace
Delete to line startCtrl+U
Delete to line endCtrl+K
SubmitEnter
Stop turn / clear inputEsc
Exit (2× to confirm)Ctrl+C
Attach imageCtrl+G
AutocompleteTab

Esc stops the current turn while generating; clears the input otherwise.

Ctrl+C — press twice within 2s to exit. Does not stop a turn — use Esc for that.

During a Running Agent

While the agent is executing, only these commands are accepted:

All other input is rejected with a notice. Use Esc to stop the current turn (clears input when idle).