Stage C controlled-failure release: open evaluation data and verification code for organoid intelligence

Machine-readable release package for a controlled-failure result in organoid-intelligence evaluation. A reservoir-computing classifier reached 1.000 test accuracy distinguishing baseline vs. post-drug organoid MEA recordings (Trujillo et al.), but four independent session-only null controls also reached up to 1.000, showing the pipeline separates recording sessions rather than drug effect. The package contains the run tables (runs.csv, blocks.csv), config with SHA-256 hashes of the raw source files, environment spec, a verification script (run: python verify.py — all 8 checks pass), and license. Code is MIT licensed (see LICENSE); data tables are CC-BY-4.0. Living version of this project: https://github.com/ontiverosreve/organoid-evaluation-standards