runlisted

# Run Skill > Harness names in these examples are anchors, not prescriptions — substitute > from the harnesses installed on this machine (`agents view`). Spreading runs > across installed harnesses is the default; the `teams-roster-guard` enforces > it for team rosters. Dispatch a single agent for a one-off task. `agents run` is the fundamental command for interactive sessions and headless automation across Claude, Codex, Gemini, Cursor, OpenCode, and OpenClaw. ## Headless vs interactive - **Prompt provided** → headless. Pipes stdout, no TTY, exits when the agent finishes. - **Prompt omitted** → interactive. Launches the agent's TUI with full stdio inheritance. ```bash # Interactive (TUI) agents run claude # Headless one-shot agents run grok "summarize recent git commits" ``` ## Modes Permission mode controls what the agent can do. | Mode | What it allows | |------|----------------| | `plan` (default) | Read-only where supported. Unsupported harnesses warn and degrade to their safest native mode (usually writable `edit`); headless Kimi rejects `plan`. | | `edit` | Read + write files; prompts for shell / risky operations | | `auto` | Harness-native automatic approval: Claude/Copilot use the smart classifier, which still prompts for risky operations; Codex uses `approval_policy=never` over `edit`'s sandbox and never prompts at all; Droid uses `--auto high`; Kimi uses `--auto` interactively, while headless `-p` already auto-approves and emits no mode flag. | | `skip` |