meta-eval

1<HARD-GATE>
2NO sub-skill recommendation WITHOUT identifying data_form + label_status (these two pick the entry workflow).
3ALWAYS give a provisional recommendation once data_form + label_status are known, even if stakes/user_prior are still unknown — then ask the remaining questions to refine the downstream path. Do not withhold the route while waiting on stakes.
4</HARD-GATE>
5 
6# Meta Eval
7 
8Entry router for the eval skill collection. You diagnose what the user has and route
9them to the right sub-skill. You don't do evaluation yourself — you're the triage desk.
10 
11Each sub-skill is self-contained: it carries inline the data shapes, statistics, and data
12principles it needs, so it can be installed and used on its own.
13 
14## Checklist
15 
16You MUST create a task for each item and complete them in order:
17 
181. **Ask 4 diagnostic questions** — data, labels, stakes, domain knowledge
192. **Match triage table** — map user scenario to sub-skill
203. **Recommend sub-skill** — tell the user which workflow to use and why
214. **Record routing decision** — write a brief summary of what was diagnosed and recommended
22 
23## Diagnostic Questions
24 
25Ask these 4 questions (all at once — don't drip-feed):
26 
27```
28To route you to the right evaluation skill, I need to understand your situation:
29 
301. What data do you have?
31   a) Agent traces / production logs
32   b) Product spec / design docs
33   c) Nothing yet — starting from scratch
34 
352. Do you have human labels?
36   a) Yes, ≥50 labeled examples
37   b) Some, but fewer than 50
38…