| name | run-economics |
| description | Use when choosing between cheap probes, worker runs, expensive ruler runs, or long keepalive waits. |
Run Economics
Match the instrument to the question.
Mechanism
Use cheap probes when the question is "does this mechanism fire?"
Examples:
- hook blocks stale wait
- cue reaches worker
- small parser case passes
- one narrow acceptance test catches the bug
Ruler
Use premium runs when the question is "does this meet the bar?"
Examples:
- release readiness
- final quality score
- customer-facing proof
- hard architectural judgment
Cache
For short waits, preserve the prompt cache with a sleep-tick if no inline work
remains. For long waits, let the cache die once. Do not ping through multi-hour
idle windows by default.
Ledger
For scored or judged runs, record provider, model or lane, scenario, method,
score, and proof boundary.