flynn/docs/plans/artifacts/phase0_baseline_live_backend_drift_2026-02-27.json

{
  "generated_at": "2026-02-27T17:36:02.803Z",
  "artifacts_dir": "/home/will/lab/flynn/docs/plans/artifacts",
  "backends": [
    "pi_embedded",
    "native"
  ],
  "report_tag": "2026-02-27",
  "max_age_hours": 36,
  "thresholds": {
    "requireBaselineHistory": false,
    "minCandidateSampledEvents": 10,
    "maxSampledEventsDropPct": 80,
    "maxRunOutcomesDropPct": 80,
    "maxCompletionRateDropPp": 35,
    "maxCancelRateIncreasePp": 25,
    "maxErrorRateIncreasePp": 25,
    "maxCancelLatencyP95IncreaseMs": 6000
  },
  "overall_pass": true,
  "reports": {
    "summary_json_out": "/home/will/lab/flynn/docs/plans/artifacts/phase0_baseline_live_backend_drift_2026-02-27.json",
    "summary_md_out": "/home/will/lab/flynn/docs/plans/artifacts/phase0_baseline_live_backend_drift_2026-02-27.md"
  },
  "results": [
    {
      "backend": "pi_embedded",
      "pass": true,
      "candidate": {
        "tag": "2026-02-27",
        "path": "/home/will/lab/flynn/docs/plans/artifacts/phase0_baseline_live_backend_pi_embedded_2026-02-27.json",
        "generated_at": "2026-02-27T17:36:02.214Z"
      },
      "baseline": null,
      "comparison": {
        "baseline": null,
        "candidate": {
          "source_event_count": 115,
          "sampled_event_count": 59,
          "run_total_outcomes": 26,
          "completion_rate_pct": 100,
          "cancel_rate_pct": 0,
          "error_rate_pct": 0,
          "cancel_latency_p95_ms": null,
          "reaction_match_rate_pct": 0,
          "reaction_skip_rate_pct": 100
        },
        "deltas": {
          "sampled_event_count_pct": null,
          "run_total_outcomes_pct": null,
          "completion_rate_pp": null,
          "cancel_rate_pp": null,
          "error_rate_pp": null,
          "cancel_latency_p95_ms": null,
          "reaction_match_rate_pp": null,
          "reaction_skip_rate_pp": null
        }
      },
      "freshness": {
        "enabled": true,
        "pass": true,
        "actual_age_hours": 0,
        "threshold_hours": 36
      },
      "drift_gate": {
        "pass": true,
        "criteria": [
          {
            "criterion": "candidate_sampled_events",
            "pass": true,
            "actual": "59",
            "threshold": ">= 10"
          },
          {
            "criterion": "sampled_events_drop_pct",
            "pass": true,
            "actual": "n/a",
            "threshold": "<= 80"
          },
          {
            "criterion": "run_outcomes_drop_pct",
            "pass": true,
            "actual": "n/a",
            "threshold": "<= 80"
          },
          {
            "criterion": "completion_rate_drop_pp",
            "pass": true,
            "actual": "n/a",
            "threshold": "<= 35"
          },
          {
            "criterion": "cancel_rate_increase_pp",
            "pass": true,
            "actual": "n/a",
            "threshold": "<= 25"
          },
          {
            "criterion": "error_rate_increase_pp",
            "pass": true,
            "actual": "n/a",
            "threshold": "<= 25"
          },
          {
            "criterion": "cancel_latency_p95_increase_ms",
            "pass": true,
            "actual": "n/a",
            "threshold": "<= 6000"
          }
        ]
      }
    },
    {
      "backend": "native",
      "pass": true,
      "candidate": {
        "tag": "2026-02-27",
        "path": "/home/will/lab/flynn/docs/plans/artifacts/phase0_baseline_live_backend_native_2026-02-27.json",
        "generated_at": "2026-02-27T17:36:02.514Z"
      },
      "baseline": null,
      "comparison": {
        "baseline": null,
        "candidate": {
          "source_event_count": 115,
          "sampled_event_count": 15,
          "run_total_outcomes": 2,
          "completion_rate_pct": 100,
          "cancel_rate_pct": 0,
          "error_rate_pct": 0,
          "cancel_latency_p95_ms": null,
          "reaction_match_rate_pct": 0,
          "reaction_skip_rate_pct": 100
        },
        "deltas": {
          "sampled_event_count_pct": null,
          "run_total_outcomes_pct": null,
          "completion_rate_pp": null,
          "cancel_rate_pp": null,
          "error_rate_pp": null,
          "cancel_latency_p95_ms": null,
          "reaction_match_rate_pp": null,
          "reaction_skip_rate_pp": null
        }
      },
      "freshness": {
        "enabled": true,
        "pass": true,
        "actual_age_hours": 0,
        "threshold_hours": 36
      },
      "drift_gate": {
        "pass": true,
        "criteria": [
          {
            "criterion": "candidate_sampled_events",
            "pass": true,
            "actual": "15",
            "threshold": ">= 10"
          },
          {
            "criterion": "sampled_events_drop_pct",
            "pass": true,
            "actual": "n/a",
            "threshold": "<= 80"
          },
          {
            "criterion": "run_outcomes_drop_pct",
            "pass": true,
            "actual": "n/a",
            "threshold": "<= 80"
          },
          {
            "criterion": "completion_rate_drop_pp",
            "pass": true,
            "actual": "n/a",
            "threshold": "<= 35"
          },
          {
            "criterion": "cancel_rate_increase_pp",
            "pass": true,
            "actual": "n/a",
            "threshold": "<= 25"
          },
          {
            "criterion": "error_rate_increase_pp",
            "pass": true,
            "actual": "n/a",
            "threshold": "<= 25"
          },
          {
            "criterion": "cancel_latency_p95_increase_ms",
            "pass": true,
            "actual": "n/a",
            "threshold": "<= 6000"
          }
        ]
      }
    }
  ]
}