Persistently loops until task is fully complete. Used for "until done", "complete it", "dont stop" requests.
8 stars
0 votes
0 copies
1 view
Added February 7, 2026
ai-agentstypescript
Security analysis
A100/100
Scanned February 12, 2026
$npx -y skills add seokan-jeong/team-shinchan --skill ralph --agent claude-code
Installs into .claude/skills of the current project.
Are you the author of Ralph?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/seokan-jeong-ralph)
---
name: team-shinchan:ralph
description: Persistently loops until task is fully complete. Used for "until done", "complete it", "dont stop" requests.
user-invocable: true
---
# EXECUTE IMMEDIATELY
**Do not read further. Execute this Task NOW:**
```typescript
Task(
subagent_type="team-shinchan:kazama",
model="opus",
prompt=`/team-shinchan:ralph has been invoked.
## Persistent Execution Until Complete Mode
Don't stop until complete:
1. Check TODO list
2. Execute next task (delegate to appropriate agent)
3. Verify results
4. On failure → Analyze cause → Retry
5. On success → Next task
6. All tasks complete → Action Kamen final verification
7. Verification fails → Fix and re-verify
## Completion Conditions
Complete only when all conditions met:
- All TODO list items completed
- Build/tests pass
- Action Kamen review approved
**If conditions not met, automatically continue!**
User request: ${args || '(Task to complete)'}
`
)
```
**STOP HERE. The above Task handles everything.**