{
  "ocm": "1",
  "id": "claudianus-superliora-loop",
  "kind": "skill",
  "name": "loop",
  "description": "Stateless agent loop: `runTurn`, tool-call batching, scheduling, guards.",
  "publisher": "claudianus",
  "version": "1.0.0",
  "capabilities": {
    "domains": [
      "general"
    ],
    "tags": [
      "skill-md",
      "github-agents"
    ],
    "languages": [
      "en"
    ]
  },
  "quality_prior": 0.6,
  "examples": [
    "Stateless agent loop: `runTurn`, tool-call batching, scheduling, guards."
  ],
  "primary": false,
  "metadata": {
    "source": {
      "provider": "github-agents",
      "repository": "https://github.com/claudianus/superliora",
      "path": "packages/agent-core/src/loop/AGENTS.md",
      "ref": "62fc4a0c70f60a2e6a0852d5ad58c23114e9c875",
      "url": "https://github.com/claudianus/superliora/blob/62fc4a0c70f60a2e6a0852d5ad58c23114e9c875/packages/agent-core/src/loop/AGENTS.md",
      "key": "claudianus/superliora/packages/agent-core/src/loop/AGENTS.md"
    }
  },
  "instructions": "# `src/loop/`\n\nStateless agent loop: `runTurn`, tool-call batching, scheduling, guards.\n\n## Ownership\n\n- Must stay free of host-layer imports (`session/`, RPC transport, permission UI). Hosts adapt into loop contracts.\n- Guard/circuit-breaker state lives in `tool-call-guards.ts` (and siblings) — keep `tool-call.ts` as orchestration.\n\n## Tests\n\n`test/loop/` or tools harness tests that exercise the loop. Prefer pure unit tests for guards.",
  "cost": {
    "context_tokens": 110
  }
}