What problem does it solve? Calibration campaigns run many rounds, and without a cross-round record a later round can re-propose parameters that earlier rounds already tested and refuted, or compare rounds that ran on different model binaries. This Skill builds the cross-round parameter ledger, mechanism ledger, and model-evolution comparability verdict that every round report requires, including the first round. ## Core Features & Use Cases - Cross-Round Parameter Ledger: One row per parameter ever in play, one column per round, with statuses like calibrated, refuted with mechanism, baked into base, or bounds provisional, sourced from parameter lists, round configs, state files, and investigation reports. - Mechanism and Model-Evolution Ledgers: Records what the campaign has learned about the system (one row per established mechanism with evidence) and verifies which model binary each round ran via git ancestry checks before claiming comparability. - Performance and Sensitivity Figures: Regenerates multiround top-50 biomass overlays and Morris mu* sensitivity comparisons per model backend, with documented footguns such as screening contamination, stale indices files, and partial-ensemble caveats. - Use Case: After round R3 of an EcoSIM calibration completes, run this Skill to produce the R1-R3 parameter ledger, confirm R3's binary ancestry, and refresh the multiround comparison figures before writing the round report. ## Quick Start Ask the agent to compare all completed calibration rounds for the current case and regenerate the cross-round parameter ledger and multiround figures before the round report.
