eval-magic 0.5.0

One-stop CLI for running skill evals — measure whether an agent skill actually shifts behavior.
Documentation
{
  "mode": "new-skill",
  "conditions": [
    {
      "name": "with_skill",
      "skill_path": "/tmp/eval/skill-dir/mr-review/SKILL.md",
      "staged_skill_slug": "slow-powers-eval-1-with-skill-mr-review"
    },
    {
      "name": "without_skill",
      "skill_path": null,
      "staged_skill_slug": null
    }
  ],
  "timestamp": "2026-07-14T04:29:09.075Z",
  "harness": "opencode",
  "run_nonce": "mrk5j619-276678",
  "runs": 1
}