/mode

description: Control Governor. Show or set mode (on/off), allow full output for the next diagnostic step, or apply strict governance to the current task.
disable-model-invocation: false

Governor Mode

Read the first word of `$ARGUMENTS` and act accordingly. If `$ARGUMENTS` is empty, report the current mode.

No arguments — report current mode

python3 "${CLAUDE_PLUGIN_ROOT}/scripts/governor.py" mode status

`on` — enable compact professional mode

python3 "${CLAUDE_PLUGIN_ROOT}/scripts/governor.py" mode compact

Confirm briefly: compact mode is on. It reinforces dense professional output on every prompt while hooks are active.

`off` — disable compact response mode

python3 "${CLAUDE_PLUGIN_ROOT}/scripts/governor.py" mode off

Confirm briefly: compact response mode is off. Tool filtering and telemetry hooks still run if configured.

`full` — full tool output for the next step

python3 "${CLAUDE_PLUGIN_ROOT}/scripts/governor.py" full

This sets a one-tool-call override; it does not disable hooks globally.

• Prefer focused commands over broad reruns.
• Do not summarize away critical lines.
• If output is enormous, ask before rerunning with full logs.

Mention that `GOVERNOR_FULL=1 <command>` does the same thing inline, and is immune to parallel-call cancellation.

`strict` — stricter governance for the current task

Guidance only: unlike `on`, `off`, and `full`, this stores no state. It shapes your behavior for the current task and nothing else. Do not carry it into unrelated future tasks, and do not report it as a persisted mode.

• Require an implementation contract before broad edits.
• Run a drift check after edit phases.
• Avoid broad scans unless justified.
• Stop and report when requirements are ambiguous.

Notes

Plain language works too and needs no command: "turn off governor" or "enable governor" is detected by the prompt hook.