---
description: Enable, disable, or show all PRD Plugin hook behavior across Claude, Codex, and OpenCode
argument-hint: "status | enable | disable"
allowed-tools: Bash(python:*)
---

Run the PRD Plugin hooks helper, then report the result plainly.

The argument is: `$ARGUMENTS` (one of `status`, `enable`, `disable`; default to
`status` if empty).

Run `prd_hooks.py` with that action — downstream path if it exists, else hub path:

- downstream: `python .prd_plugin/scripts/prd_hooks.py $ARGUMENTS`
- hub: `python scripts/prd_hooks.py $ARGUMENTS`

Then tell the user whether the unified `hooks.enabled` master switch is on or
off. Host wiring stays installed; the dispatcher reads the switch on every
event, so the change applies to Claude, Codex, and OpenCode. The hooks
are the SessionStart/UserPromptSubmit nudge, the Skill-usage logger, the Stop
auto-report, configurable **reflection questions**, the autonomous run-until-done
**stop-guard**, and the PreToolUse
**pre-commit gate** (blocks `git commit` when `prd_gate` finds errors — opt-in via
`automation.precommit_gate: true`). Note that
`disable` turns off **all** behavior together without rewriting host manifests;
to disable only the run-until-done behavior, set
`automation.autonomous_run_until_done: false` instead. Do not do anything else.
To tune latency or disable only one behavior, use `/prd-config describe`,
`effective`, `enable`, or `disable`. To disable only reflection, set
`reflection.enabled: false` (or run
`/prd-config set reflection.enabled off`).