Slash commands

Type / in the composer to search available commands. Built-ins can change the app or run directly; dynamic entries can send an expanded prompt.

Built-in commands

Command	Behavior
/clear	Start a fresh chat for a new task boundary
/compact	Summarize older messages to free context space; trailing text can guide compaction
/marketplace	Browse and manage skills, MCP servers, and packages
/settings	Open application settings
/create-rule	Create a workspace rule under .vyotiq/rules/
/create-skill	Create under .vyotiq/skills/; add personal for personal scope
/help	Send a generated list of currently available commands
/undo	Restore files from the last agent write checkpoint for this run
/ask	Switch to Ask mode
/plan	Switch to Plan mode
/agent	Switch to Agent mode
/harness-review	Mine recent run receipts into a harness proposal draft
/harness-apply	Confirm and apply the latest or named harness proposal
/goal	Set a long-lived objective for this chat (/goal pause, resume, complete)
/loop	Repeat a prompt on an interval (/loop 30s check CI; /loop stop)

There are 15 built-in slash commands.

Dynamic command sources

The menu can also include:

• enabled local or Marketplace skills;
• tools from connected MCP servers;
• workspace rule files (the command opens the rule);
• command markdown files under .vyotiq/commands/ or .cursor/commands/.

Availability can therefore change when you switch workspaces, connect an MCP server, create a skill, or edit rules. /help is generated from commands that are ready at that moment and limits its output to 40 entries.

Syntax

Select a menu item or type its trigger and optional trailing text. For example:

/compact Keep the accepted API decision and the failing test name.

Client commands such as /settings open a view instead of sending text to the model. Mode commands retain the current chat. /clear creates the fresh task boundary.

Slash commands do not bypass mode policy, tool approval, or confirmation steps. /harness-apply, for example, remains human-confirmed.