# Loop Worker

> Run Ruflo background workers using Claude Code native /loop scheduling

- Skill: `ruvnet/loop-worker-2` (Agent Skill)
- Install (CLI): `npx get-skill add ruvnet/loop-worker-2`
- Raw SKILL.md: https://api.skillmd.com/api/skills/ruvnet/loop-worker-2/raw
- Safety review: pending (external: skill-scanner PASS, skillspector PASS)
- Works with: Claude Code, Claude.ai, OpenAI Codex
- Category: Coding & Dev Tools
- Author: ruvnet (https://skillmd.com/u/ruvnet)
- Updated: 2026-11-11
- Page: https://skillmd.com/skills/ruvnet/loop-worker-2

---

Use `/loop` for in-session periodic workers. Pick the delay from what you are waiting for:
- With the 5-minute prompt-cache TTL, `min(270, cache_ttl * 0.9)` = 270s keeps the cache warm (ruflo-autopilot ADR-0001 owns that heartbeat).
- If the `ScheduleWakeup` tool description says this session uses the 1-hour TTL, a 270s wake buys nothing; poll at 1200-1800s unless the worker finishes faster.

Dispatch a worker via MCP: `mcp__plugin_ruflo-core_ruflo__hooks_worker-dispatch({ trigger: "WORKER_NAME" })`. It runs in the background by default (`background: true`), so check on it with `hooks_worker-status` instead of waiting inline.

Then schedule the next iteration. `ScheduleWakeup` needs the loop prompt back so the next firing re-enters the loop:
`ScheduleWakeup({ delaySeconds: 270, reason: "next WORKER_NAME iteration", prompt: "/loop <the same /loop input>", noop: false })`

Stop the loop when the worker reports nothing new twice in a row: `ScheduleWakeup({ stop: true })`.

### Available Workers (12, per ADR-0001 and `/ruflo-loop`)

| Worker | Priority | Interval | Description |
|--------|----------|----------|-------------|
| audit | critical | 270s | Security analysis |
| optimize | high | 270s | Performance optimization |
| predict | normal | 270s | Predictive preloading |
| testgaps | normal | 270s | Test coverage analysis |
| ultralearn | normal | 600s | Deep knowledge acquisition |
| deepdive | normal | 600s | Deep code analysis |
| map | normal | 600s | Codebase mapping |
| document | normal | 600s | Auto-documentation |
| refactor | normal | 600s | Refactoring suggestions |
| benchmark | normal | 600s | Performance benchmarking |
| consolidate | low | 600s | Memory consolidation |
| preload | low | 270s | Resource preloading |

The intervals are suggested cadences, not enforced by the dispatcher: 270s keeps the prompt cache warm, and the heavier analysis workers run at 600s because one iteration rarely finishes sooner.

Work you start in the background (a build, a long job) finishes with its own completion notification. Do not poll it with short `ScheduleWakeup` calls; use a long fallback delay (1200s or more) so the loop survives a hung job.

MCP tool responses include `[LOOP_SUGGESTION]` and `[CRON_SUGGESTION]` hints -- follow them.

