From auto-audit
Halts the autonomous audit loop, drops the active-repo pointer so ticks become no-ops, and instructs to press Esc to cancel the running /loop.
How this skill is triggered — by the user, by Claude, or both
Slash command
/auto-audit:stop [slug][slug]This skill is limited to the following tools:
The summary Claude sees in its skill listing — used to decide when to auto-load this skill
```bash
source "${CLAUDE_PLUGIN_ROOT}/scripts/lib/common.sh"
# If a slug is given and it matches the active pointer, drop the pointer.
# If a slug is given but a different one is active, leave the active
# pointer alone (this command's intent is "stop the audit on <slug>",
# which when <slug> is not active is already true). The other repo's
# state is on disk regardless and reachable via --slug arguments.
ARG_SLUG="$1"
if [ -f "$AUTO_AUDIT_DATA/active.json" ]; then
ACTIVE="$(jq -r .slug "$AUTO_AUDIT_DATA/active.json" 2>/dev/null || echo '')"
if [ -z "$ARG_SLUG" ] || [ "$ARG_SLUG" = "$ACTIVE" ]; then
mv "$AUTO_AUDIT_DATA/active.json" "$AUTO_AUDIT_DATA/active.json.stopped.$(date +%s)"
echo "active-repo pointer removed (was '$ACTIVE'); ticks are now no-ops."
else
echo "stop requested for '$ARG_SLUG' but active is '$ACTIVE' — active pointer left in place."
echo "the '$ARG_SLUG' workspace state is preserved at $AUTO_AUDIT_DATA/repos/$ARG_SLUG/."
fi
else
echo "no active audit."
fi
echo
echo "to cancel the running loop, press Esc in the main chat."
echo "state is preserved at $AUTO_AUDIT_DATA/repos/ — you can resume later with /auto-audit:resume [slug]."
npx claudepluginhub wrxck/claude-plugins --plugin auto-auditResumes a stopped autonomous audit session without re-initializing the workspace. Recovers findings stuck mid-tick from interrupted loops.
Cancels active Claude Code loops by checking/removing .loophaus/state.json or legacy .claude/ralph-loop.local.md files and reporting stopped iteration.
Manually runs consensus-loop audits reviewing pending trigger_tag items in watch files. Use to trigger audits without editing files, re-run failures, test prompts, with dry-run, auto-fix, and model options.