Loading skill
Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Mid-run failure recovery for disease-market-sizing-orchestration. Inspects .cache/<slug>/ to determine which phase last completed and which phase to run next. Use when orchestration was interrupted and user wants to continue without restarting from phase 1.
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 1560% | 0% |
| case-02 | ✗→✓ | ▲ Improved | 445% | 0% |
| case-05 | ✗→✓ | ▲ Improved | -2% | 0% |
| case-08 | ✗→✓ | ▲ Improved | -26% | 0% |
| case-10 | ✗→✓ | ▲ Improved | -57% | 0% |
Detects partial state of an interrupted disease-market-sizing-orchestration run by scanning the per-disease cache directory.
detect_partial_state(slug_dir: Path) -> PhaseState — returns last_completed_phase + next_phase + reason. resume_from(slug_dir: Path) -> str — human-readable resume message.
| Phase complete | Signal in .cache/<slug>/ | |---|---| | 1 (contract) | contract.json exists | | 2 (staging) | staging.json exists | | 3 (recall) | sources/ non-empty | | 4 (compose+audit) | both citations_index.json and verification_report.json exist with verdict: "ok" | | 5 (bundle) | output/delivery_manifest.json exists |
Always called by disease-market-sizing-orchestration.run() at the very start to decide whether to start fresh or continue from a partial run.
Other measured skills in the registry, with their headline benchmark lift.