Commands

OpenCorvus provides a slash-command system for session prompts, including ACP session/prompt, with a /command prefix.

Built-in slash commands

Command	Purpose
/init	Initialize an AGENTS.md from the current repo
/share	Generate a share link for the current session
/undo	Roll back the last agent change when snapshots are explicitly enabled
/help	Show the command map

The exact set is version-dependent — /help is authoritative.

Custom commands

Define commands in config under command:

{
  "command": {
    "test": {
      "template": "Run the full test suite with coverage report and show any failures.",
      "description": "Run tests with coverage",
      "agent": "build",
      "model": "anthropic/claude-haiku-4-5",
    },
    "component": {
      "template": "Create a new React component named $ARGUMENTS with TypeScript.",
      "description": "Create a new component",
    },
  },
}

$ARGUMENTS is replaced with whatever the user types after the command name.

Markdown-defined commands

You can also define commands as markdown files under <runtime-root>/config/commands/<name>.md or .opencorvus/commands/<name>.md. The frontmatter holds metadata (description, agent, model); the body is the prompt template. The runtime root follows Environment variables.

Permissions in commands

Custom commands run through the same permission system as direct prompts. If your command’s template ends up triggering bash or write, Permissions still apply.

Snapshots & /undo

/undo requires snapshot: true in config. Snapshot capture is disabled by default; without that explicit setting, /undo is a no-op.