{
  "schema_version": "1.0",
  "updated_at": "2026-10-12",
  "scope": "Small curated guide examples, separate from the full bibliography. No ranking or measured results.",
  "unknown_policy": "Missing or null values mean unknown, never zero, free, permitted, or unavailable.",
  "items": [
    {
      "id": "appworld",
      "name": "AppWorld",
      "execution_setting": "stateful-software-simulation",
      "tags": [
        "simulation",
        "state-verification",
        "app-workflows"
      ],
      "source_ids": [
        "appworld"
      ],
      "summary": "Stateful application task environments with state-based evaluation.",
      "summary_zh": "带状态评测的有状态应用任务环境。",
      "limitations": [
        "Application tasks and verification do not establish fidelity to arbitrary real services."
      ],
      "source_urls": [
        "https://github.com/StonyBrookNLP/appworld"
      ],
      "evidence": "primary-source-description",
      "independently_reproduced": false,
      "checked_at": "2026-10-12"
    },
    {
      "id": "toolsandbox",
      "name": "ToolSandbox",
      "execution_setting": "stateful-tool-simulation",
      "tags": [
        "simulation",
        "state-verification",
        "milestones"
      ],
      "source_ids": [
        "toolsandbox"
      ],
      "summary": "Stateful tool scenarios evaluated using ordered milestones.",
      "summary_zh": "使用有序里程碑评测的有状态工具场景。",
      "limitations": [
        "Scenario and milestone coverage bound the conclusions."
      ],
      "source_urls": [
        "https://github.com/apple-aiml-research/ToolSandbox"
      ],
      "evidence": "primary-source-description",
      "independently_reproduced": false,
      "checked_at": "2026-10-12"
    },
    {
      "id": "envscaler",
      "name": "EnvScaler",
      "execution_setting": "generated-executable-simulation",
      "tags": [
        "simulation",
        "environment-generation",
        "programmatic"
      ],
      "source_ids": [
        "envscaler"
      ],
      "summary": "Generation of executable stateful agent environments.",
      "summary_zh": "生成可执行的有状态 Agent 环境。",
      "limitations": [
        "Generation requires separate behavioral and verifier validation."
      ],
      "source_urls": [
        "https://github.com/RUC-NLPIR/EnvScaler"
      ],
      "evidence": "primary-source-description",
      "independently_reproduced": false,
      "checked_at": "2026-10-12"
    },
    {
      "id": "mirrorapi",
      "name": "StableToolBench-MirrorAPI",
      "execution_setting": "api-behavior-simulation",
      "tags": [
        "simulation",
        "api-mirrors",
        "fidelity"
      ],
      "source_ids": [
        "mirrorapi"
      ],
      "summary": "A published approach to modeling real API environments.",
      "summary_zh": "对真实 API 环境建模的已发表方案。",
      "limitations": [
        "Reported API coverage is not evidence of complete real-service state fidelity."
      ],
      "source_urls": [
        "https://aclanthology.org/2025.findings-acl.273/"
      ],
      "evidence": "primary-source-description",
      "independently_reproduced": false,
      "checked_at": "2026-10-12"
    }
  ]
}