What problem does it solve? EcoSIM runs can complete successfully yet produce dead, stunted, or botanically wrong plants because the per-PFT trait VALUES in the pft_file_in NC table are physiologically inappropriate. This Skill diagnoses those value problems by running the EcoSIM developer's deterministic checker against the run's plant_trait.*.desc dump, then guides refinement of the NC input and re-verification by run. ## Core Features & Use Cases - Deterministic trait checking: Runs check_plant_trait_desc.py on a plant_trait.<year>.desc dump to flag range, fraction, sign, N/P magnitude, type-flag, and cross-parameter issues ranked ERROR to WARN. - Optional web-evidence layer: Derives literature-based trait ranges per PFT via the ecosim-trait-deriver skill and re-checks with --web-evidence for borderline magnitudes. - Refine and re-verify loop: Maps flagged .desc codes back to NC pft_file_in variables, patches values with netCDF4, re-runs on the same binary, and confirms the plant actually grows (peak SHOOT_C_pft), not just that the checker passes. - Use Case: An EcoSIM BioCON run finishes and writes output tapes, but the plant regrows smaller every spring. Use this Skill to sanity-check the trait values, test hypotheses like the RCS sign convention, and determine whether the cause is parameter values or a deeper carbon-balance issue requiring recalibration. ## Quick Start Ask the agent to sanity-check the plant trait values from my EcoSIM run's plant_trait.2000.desc file and tell me which pft input parameters look physiologically wrong.
