{
  "format": "agl-live-smoke-v1",
  "verified_at": "2026-11-04",
  "status": "PASS",
  "model": "deepseek-flash",
  "actual_provider": true,
  "actual_docker": true,
  "task_fixture": "sum of integers 1 through 10, print only the integer",
  "code": "print(sum(range(1, 11)))",
  "stdout": "55\n",
  "exit_code": 0,
  "cleanup_confirmed": true,
  "effect_count": 1,
  "reviewer_verdict": "approve",
  "approval": {
    "automated_synthetic": true,
    "human_review_claimed": false,
    "scope": "arithmetic fixture only, AST allowlist before approval"
  },
  "usage_source": "authoritative provider usage persisted in agent_calls; no usage estimated",
  "calls": [
    {
      "role": "researcher",
      "input_tokens": 206,
      "output_tokens": 128,
      "usage_known": true,
      "booked_micro_usd": 216
    },
    {
      "role": "coder",
      "input_tokens": 334,
      "output_tokens": 21,
      "usage_known": true,
      "booked_micro_usd": 126
    },
    {
      "role": "reviewer",
      "input_tokens": 229,
      "output_tokens": 44,
      "usage_known": true,
      "booked_micro_usd": 122
    }
  ],
  "booked_micro_usd": 464,
  "reserved_micro_usd": 0,
  "rate_basis": {
    "model": "deepseek-flash",
    "input_micro_usd_per_million": 300000,
    "output_micro_usd_per_million": 1200000,
    "basis": "conservative peak cache-miss upper bound applied to authoritative usage tokens; not an invoice",
    "source": "https://api-docs.deepseek.com/quick_start/pricing/",
    "verified_at": "2026-11-04",
    "cache_discount_accounted": false,
    "off_peak_discount_accounted": false
  },
  "limits": [
    "This is one arithmetic smoke, not a general model-quality evaluation.",
    "Booked cost applies a conservative peak cache-miss tariff, not a provider invoice.",
    "This evidence does not verify human approval or delivery in real Slack/Telegram channels.",
    "The smoke retained traces locally; it does not verify Langfuse cloud receipt."
  ]
}