T-mn82bigk.json 1.16 KB

Edit Raw Blame History



1

2

3

4

5

6

7

8

9

10

11

12

{
  "taskId": "T-mn82bigk",
  "from": "innovator",
  "to": "digester",
  "task": "【DIGEST任务 - researcher深度研究技能】<br><br>输入路径：<br>- 调研产出: ../../workspace/hunter/researcher/knowhow_raw.md<br>- 爬取配置: ../../workspace/hunter/researcher/crawl_config.json<br>- 种子数据: ../../workspace/hunter/researcher/seed_data.json<br><br>输出路径：../../workspace/digester/researcher/<br><br>任务描述：<br>将 Hunter 的原始调研产出进行结构化消化，生成：<br>1. knowhow.md - 结构化知识体系（研究方法论、剧本创作流程、数据源清单）<br>2. benchmark.md - 评测标准（输入输出格式、能力维度、评分规则、pass_score=90）<br>3. train_dataset.json - 训练数据集（30+条精选样本，含输入输出对）<br><br>验收标准：<br>- knowhow.md 包含 3+ 研究维度、剧本创作方法论<br>- benchmark.md 定义清晰的评分标准和 pass_score=90<br>- train_dataset.json 提供可直接用于训练的样本",
  "priority": "P1",
  "goalId": null,
  "nodeId": null,
  "executionConvId": "task-T-mn82bigk",
  "createdAt": "2026-03-26T22:47:13.232Z",
  "status": "dispatched"
}