| name | team-shinchan:fierce-ralph |
| description | Deterministic loop-until-done for high-stakes long-running tasks — a worker/verifier loop the script bounds by iteration cap, token budget, and stagnation, closed by an Action-Kamen gate. Opt-in main-loop Workflow tier. |
| user-invocable | false |
EXECUTE IMMEDIATELY
Fierce ralph is a main-loop Workflow whose loop condition is owned by the SCRIPT, not narrated by an agent: a worker does the next unit of work, a verifier independently checks progress + completion against the real repo, and it repeats until done — bounded by a hard iteration cap, a token budget, and a stagnation limit, then closed by an Action-Kamen final gate. The Task path (team-shinchan:ralph, Kazama's narrated boulder loop) can describe "don't stop until done"; this tier enforces it deterministically. Use ONLY for high-stakes, genuinely long-running work where a stalled or prematurely-stopped loop is costly. The cheap default stays team-shinchan:ralph.
Step 0: Validate + opt-in
- This skill calls the Workflow tool (main-loop only). Invoking
/team-shinchan:fierce-ralph IS the opt-in. Never delegate to a subagent — workflow() throws inside a Task child.
- If args lack a clear goal: ask the user for the goal (and success criteria) and STOP. If > 2000 chars: truncate + warn.
- The loop spawns file-editing agents and is token-heavy — recommend the user set a budget (e.g. append a
+500k-style target) so the loop is bounded.
Step 1: Resolve goal + personas (main loop — the script can't read files)
