{
  "ocm": "1",
  "id": "narcolepticchicken-teacherlab-lesson-11",
  "kind": "skill",
  "name": "lesson 11",
  "description": "Teach reward misspecification and constrained policy selection.",
  "publisher": "narcolepticchicken",
  "version": "1.0.0",
  "capabilities": {
    "domains": [
      "general"
    ],
    "tags": [
      "skill-md",
      "github-agents"
    ],
    "languages": [
      "en"
    ]
  },
  "quality_prior": 0.6,
  "examples": [
    "Teach reward misspecification and constrained policy selection."
  ],
  "primary": false,
  "metadata": {
    "source": {
      "provider": "github-agents",
      "repository": "https://github.com/narcolepticchicken/teacherlab",
      "path": "lesson 11/AGENTS.md",
      "ref": "7beca52851cc8439744373fec273d6c0c7e2b1b4",
      "url": "https://github.com/narcolepticchicken/teacherlab/blob/7beca52851cc8439744373fec273d6c0c7e2b1b4/lesson 11/AGENTS.md",
      "key": "narcolepticchicken/teacherlab/lesson 11/AGENTS.md"
    }
  },
  "instructions": "# Lesson 11 Agent Notes\n\nTeach reward misspecification and constrained policy selection.\n\n- Keep proxy reward and safety cost as separate signals.\n- Never treat a hard authority constraint as an ordinary tradeable penalty.\n- Report both expected reward and violation rate for every policy.\n- State that all distributions and values are invented.",
  "cost": {
    "context_tokens": 86
  }
}