{
  "schema_version": "q36.laptop-capability-plan.v1",
  "status": "fixtures_partially_prepared",
  "total_case_slots": 64,
  "benchmarks": [
    {
      "name": "HumanEval+",
      "cases": 5,
      "success_evidence": "Generated functions pass the selected task's base and extended tests in an isolated container",
      "status": "Existing outputs; execution/grading must be reconciled"
    },
    {
      "name": "MBPP+",
      "cases": 5,
      "success_evidence": "Correct results on base and extended tests, including edge cases",
      "status": "Existing outputs; execution/grading must be reconciled"
    },
    {
      "name": "LiveCodeBench",
      "cases": 9,
      "success_evidence": "Program passes task tests within sandbox limits",
      "status": "Existing outputs; execution/grading must be reconciled"
    },
    {
      "name": "Q36 bug fixes",
      "cases": 5,
      "success_evidence": "Reproduce a failing test, apply a patch, pass regression tests",
      "status": "New tasks to prepare"
    },
    {
      "name": "Q36 small app / website tasks",
      "cases": 5,
      "success_evidence": "Build succeeds; required UI behavior passes browser checks",
      "status": "New tasks to prepare"
    },
    {
      "name": "Tool choice and arguments",
      "cases": 5,
      "success_evidence": "Correct tool and schema-valid arguments; no invented tool result",
      "status": "Local smoke fixtures prepared separately"
    },
    {
      "name": "Multi-step planning",
      "cases": 5,
      "success_evidence": "Respect dependencies and reach a verifiable final state",
      "status": "New interactive fixtures to prepare"
    },
    {
      "name": "Recovery from tool errors",
      "cases": 5,
      "success_evidence": "Respond to a simulated failure, choose a valid alternative, verify recovery",
      "status": "New interactive fixtures to prepare"
    },
    {
      "name": "Memory and conflicting evidence",
      "cases": 5,
      "success_evidence": "Retrieve the correct note, cite it, distinguish updated from stale facts",
      "status": "New isolated test vault to prepare"
    },
    {
      "name": "Stopping and task completion",
      "cases": 5,
      "success_evidence": "Stop after success; report missing prerequisites instead of inventing completion",
      "status": "New interactive fixtures to prepare"
    },
    {
      "name": "Transfer to changed constraints",
      "cases": 5,
      "success_evidence": "Solve held-out variants after a changed rule or environment",
      "status": "New tasks to prepare"
    },
    {
      "name": "Q36 browser workflows",
      "cases": 5,
      "success_evidence": "Complete a task on a local test website and verify the resulting page state",
      "status": "New interactive fixtures to prepare"
    }
  ],
  "general_agi_proof": false
}