{
  "id": "test-set:terminal-bench-v1",
  "_kind": "TestSet",
  "_file": "benchmarks/test-sets/terminal-bench-v1.yaml",
  "_cluster": "benchmarks",
  "attributes": {
    "displayName": "Terminal-Bench v1",
    "benchmarkId": "benchmark:terminal-bench",
    "caseCount": 80,
    "releasedAt": "2024-10-01",
    "composition": "The v1 release of Terminal-Bench from Stanford NLP / Princeton.\nEach task is a multi-step shell scenario evaluated end-to-end in a\nDocker sandbox; success requires the agent to reach a target file\nstate via real shell commands.\n",
    "homepageUrl": "https://www.tbench.ai/",
    "description": "Canonical Terminal-Bench v1 set referenced in the original paper\nand the public leaderboard.\n"
  },
  "outgoingEdges": [
    {
      "from": "test-set:terminal-bench-v1",
      "to": "benchmark:terminal-bench",
      "kind": "belongs_to_benchmark"
    }
  ],
  "incomingEdges": []
}