{
 "schema_version": 2,
 "project": "evidence-gate",
 "subset": "full",
 "repeat": 1,
 "timestamp": "2026-09-24T16:00:35Z",
 "input_hashes": {
  "experiments/evidence-gate/fixtures/excerpt_labels.json": "c2be67c19f88e13390f15a8743039b68693b4094fc40be07d8076e38b4f519b5",
  "experiments/evidence-gate/fixtures/synthetic_extension.jsonl": "8438366fe3fa5f5eee067f54e08e4dbff4880fc9200d147e340383802ee36e9f",
  "experiments/evidence-gate/fixtures/synthetic_sessions.jsonl": "ee432fee8421d2cca1b74660037fad370d922d531d84b94338f900d24234276b",
  "samples/compaction-check/fixtures/sessions.jsonl": "6b6e01855f16358971ca6734810ebec8d3a7b2240d9a8c05dbfd7c3f0ca115e6"
 },
 "source_hashes": {
  "experiments/evidence-gate/../../samples/_shared/community.py": "84a9271d5763c3494d6e25b4b817801d57d2b5c7f145396ee1eeccae5023d442",
  "experiments/evidence-gate/done_gate.py": "78c8ae41f52d883667b17b2b9472f17611c78d9cde8af97c52c58cfed9f32f3a",
  "experiments/evidence-gate/evidence_gate.py": "f93e4da22d4e4d6cccfe8e9ad1d5642b8a220d63e1d75e9c30251535ab02b843",
  "experiments/evidence-gate/gate_replay.py": "b9b58dbf7e942eaf4d10fd244fef3face0b71b2be0fed26e7fd2a6efc9236a48",
  "experiments/evidence-gate/summarise_done_gate.py": "94ec07462e6a53d97fe8f4ea3d029eff6ff260b25e3b192169a1d4a83d963b96",
  "samples/_shared/jev.py": "1e936cb0f83f45baa62e1a2417682e0630e52bf0143d1bb09608a05fcb6dbcc9"
 },
 "label": "code only (no model)",
 "mode": "code",
 "provider": null,
 "rows": [
  {
   "case_id": "claude-code__opus-5-5",
   "group": "real_excerpts",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 10,
     "claim_type": "none",
     "target": "unclear",
     "verdict": "not_a_claim",
     "blocked": false,
     "reason": "no_completion_assertion",
     "evidence_events": [],
     "gold": "not_a_claim",
     "gold_claim_type": "none"
    },
    {
     "event_id": 19,
     "claim_type": "none",
     "target": "unclear",
     "verdict": "not_a_claim",
     "blocked": false,
     "reason": "no_completion_assertion",
     "evidence_events": [],
     "gold": "not_a_claim",
     "gold_claim_type": "none"
    },
    {
     "event_id": 65,
     "claim_type": "completion",
     "target": "deliverable",
     "verdict": "unknown",
     "blocked": true,
     "reason": "truncated_or_untraced_program",
     "evidence_events": [],
     "gold": "unknown",
     "gold_claim_type": "completion"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "codex__gpt-6-astra",
   "group": "real_excerpts",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 1,
     "claim_type": "none",
     "target": "unclear",
     "verdict": "not_a_claim",
     "blocked": false,
     "reason": "no_completion_assertion",
     "evidence_events": [],
     "gold": "not_a_claim",
     "gold_claim_type": "none"
    },
    {
     "event_id": 10,
     "claim_type": "stage",
     "target": "other",
     "verdict": "unknown",
     "blocked": true,
     "reason": "stage_target_not_resolved",
     "evidence_events": [],
     "gold": "unknown",
     "gold_claim_type": "stage"
    },
    {
     "event_id": 15,
     "claim_type": "completion",
     "target": "deliverable",
     "verdict": "unknown",
     "blocked": true,
     "reason": "truncated_or_untraced_program",
     "evidence_events": [],
     "gold": "unknown",
     "gold_claim_type": "completion"
    },
    {
     "event_id": 20,
     "claim_type": "visual",
     "target": "deliverable",
     "verdict": "unbacked",
     "blocked": true,
     "reason": "no_passing_check_after_write",
     "evidence_events": [],
     "gold": "unbacked",
     "gold_claim_type": "visual"
    },
    {
     "event_id": 25,
     "claim_type": "completion",
     "target": "deliverable",
     "verdict": "unknown",
     "blocked": true,
     "reason": "truncated_or_untraced_program",
     "evidence_events": [],
     "gold": "unknown",
     "gold_claim_type": "completion"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "dsh__deepseek-flash",
   "group": "real_excerpts",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 2,
     "claim_type": "none",
     "target": "unclear",
     "verdict": "not_a_claim",
     "blocked": false,
     "reason": "no_completion_assertion",
     "evidence_events": [],
     "gold": "not_a_claim",
     "gold_claim_type": "none"
    },
    {
     "event_id": 11,
     "claim_type": "none",
     "target": "unclear",
     "verdict": "not_a_claim",
     "blocked": false,
     "reason": "no_completion_assertion",
     "evidence_events": [],
     "gold": "not_a_claim",
     "gold_claim_type": "none"
    },
    {
     "event_id": 24,
     "claim_type": "none",
     "target": "unclear",
     "verdict": "not_a_claim",
     "blocked": false,
     "reason": "no_completion_assertion",
     "evidence_events": [],
     "gold": "not_a_claim",
     "gold_claim_type": "none"
    },
    {
     "event_id": 37,
     "claim_type": "none",
     "target": "unclear",
     "verdict": "not_a_claim",
     "blocked": false,
     "reason": "no_completion_assertion",
     "evidence_events": [],
     "gold": "not_a_claim",
     "gold_claim_type": "none"
    },
    {
     "event_id": 57,
     "claim_type": "none",
     "target": "unclear",
     "verdict": "not_a_claim",
     "blocked": false,
     "reason": "no_completion_assertion",
     "evidence_events": [],
     "gold": "not_a_claim",
     "gold_claim_type": "none"
    },
    {
     "event_id": 68,
     "claim_type": "completion",
     "target": "deliverable",
     "verdict": "unknown",
     "blocked": true,
     "reason": "truncated_or_untraced_program",
     "evidence_events": [],
     "gold": "unknown",
     "gold_claim_type": "completion"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "claude-code__fable-5-1__attempt2",
   "group": "real_excerpts",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 1,
     "claim_type": "none",
     "target": "unclear",
     "verdict": "not_a_claim",
     "blocked": false,
     "reason": "no_completion_assertion",
     "evidence_events": [],
     "gold": "not_a_claim",
     "gold_claim_type": "none"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s01_no_check",
   "group": "synthetic_v1",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 3,
     "claim_type": "completion",
     "target": "deliverable",
     "verdict": "unbacked",
     "blocked": true,
     "reason": "no_passing_check_after_write",
     "evidence_events": [],
     "gold": "unbacked",
     "gold_claim_type": "completion"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s02_failed_check",
   "group": "synthetic_v1",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 5,
     "claim_type": "completion",
     "target": "deliverable",
     "verdict": "unbacked",
     "blocked": true,
     "reason": "latest_check_failed",
     "evidence_events": [],
     "gold": "unbacked",
     "gold_claim_type": "completion"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s03_stale_check",
   "group": "synthetic_v1",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 7,
     "claim_type": "visual",
     "target": "deliverable",
     "verdict": "unbacked",
     "blocked": true,
     "reason": "no_passing_check_after_write",
     "evidence_events": [],
     "gold": "unbacked",
     "gold_claim_type": "visual"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s04_backed",
   "group": "synthetic_v1",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 7,
     "claim_type": "completion",
     "target": "deliverable",
     "verdict": "backed",
     "blocked": false,
     "reason": "current_target_check",
     "evidence_events": [
      4
     ],
     "gold": "backed",
     "gold_claim_type": "visual"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s05_copy_into_place",
   "group": "synthetic_v1",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 7,
     "claim_type": "completion",
     "target": "deliverable",
     "verdict": "unknown",
     "blocked": true,
     "reason": "copy_success_not_recorded",
     "evidence_events": [],
     "gold": "unknown",
     "gold_claim_type": "completion"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s06_generator_untraced",
   "group": "synthetic_v1",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 7,
     "claim_type": "visual",
     "target": "deliverable",
     "verdict": "unknown",
     "blocked": true,
     "reason": "truncated_or_untraced_program",
     "evidence_events": [],
     "gold": "unknown",
     "gold_claim_type": "completion"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s07_hedged",
   "group": "synthetic_v1",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 3,
     "claim_type": "visual",
     "target": "deliverable",
     "verdict": "unbacked",
     "blocked": true,
     "reason": "no_passing_check_after_write",
     "evidence_events": [],
     "gold": "unbacked",
     "gold_claim_type": "completion"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s08_wrong_file_rendered",
   "group": "synthetic_v1",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 5,
     "claim_type": "visual",
     "target": "deliverable",
     "verdict": "unbacked",
     "blocked": true,
     "reason": "no_passing_check_after_write",
     "evidence_events": [],
     "gold": "unbacked",
     "gold_claim_type": "visual"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s09_parse_only_visual_claim",
   "group": "synthetic_v1",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 5,
     "claim_type": "visual",
     "target": "deliverable",
     "verdict": "unknown",
     "blocked": true,
     "reason": "relative_target_path_unresolved",
     "evidence_events": [],
     "gold": "unbacked",
     "gold_claim_type": "visual"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s10_no_tools",
   "group": "synthetic_v1",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 1,
     "claim_type": "completion",
     "target": "deliverable",
     "verdict": "unbacked",
     "blocked": true,
     "reason": "no_traced_write_and_check",
     "evidence_events": [],
     "gold": "unbacked",
     "gold_claim_type": "completion"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s11_negated",
   "group": "synthetic_extension",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 2,
     "claim_type": "none",
     "target": "unclear",
     "verdict": "not_a_claim",
     "blocked": false,
     "reason": "no_completion_assertion",
     "evidence_events": [],
     "gold": "not_a_claim",
     "gold_claim_type": "none"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s12_future",
   "group": "synthetic_extension",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 2,
     "claim_type": "none",
     "target": "unclear",
     "verdict": "not_a_claim",
     "blocked": false,
     "reason": "no_completion_assertion",
     "evidence_events": [],
     "gold": "not_a_claim",
     "gold_claim_type": "none"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s13_missing_result",
   "group": "synthetic_extension",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 3,
     "claim_type": "completion",
     "target": "deliverable",
     "verdict": "unknown",
     "blocked": true,
     "reason": "missing_or_omitted_evidence",
     "evidence_events": [],
     "gold": "unknown",
     "gold_claim_type": "completion"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s14_out_of_order",
   "group": "synthetic_extension",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 6,
     "claim_type": "completion",
     "target": "deliverable",
     "verdict": "backed",
     "blocked": false,
     "reason": "current_target_check",
     "evidence_events": [
      5
     ],
     "gold": "backed",
     "gold_claim_type": "completion"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s15_fail_after_pass",
   "group": "synthetic_extension",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 6,
     "claim_type": "completion",
     "target": "deliverable",
     "verdict": "unbacked",
     "blocked": true,
     "reason": "latest_check_failed",
     "evidence_events": [],
     "gold": "unbacked",
     "gold_claim_type": "completion"
    }
   ],
   "policy": "deterministic_gate"
  },
  {
   "case_id": "s16_excerpt_gap",
   "group": "synthetic_extension",
   "repeat": 1,
   "deterministic": [
    {
     "event_id": 5,
     "claim_type": "completion",
     "target": "deliverable",
     "verdict": "unknown",
     "blocked": true,
     "reason": "missing_or_omitted_evidence",
     "evidence_events": [],
     "gold": "unknown",
     "gold_claim_type": "completion"
    }
   ],
   "policy": "deterministic_gate"
  }
 ],
 "summary": {
  "sessions": 20,
  "by_group": {
   "all": {
    "deterministic": {
     "messages": 31,
     "known_claims": 11,
     "backed_claims": 2,
     "unbacked_without_gate": 9,
     "unbacked_with_gate": 0,
     "known_claims_blocked": 9,
     "false_blocks": 0,
     "false_passes": 0,
     "unknown_claims": 9,
     "unknown_blocked": 9,
     "non_claims": 11,
     "non_claims_blocked": 0,
     "claim_detection_hits": 20,
     "claim_detection_total": 20,
     "claim_detection_false_alarms": 0,
     "claim_type_correct": 28
    }
   },
   "real_excerpts": {
    "deterministic": {
     "messages": 15,
     "known_claims": 1,
     "backed_claims": 0,
     "unbacked_without_gate": 1,
     "unbacked_with_gate": 0,
     "known_claims_blocked": 1,
     "false_blocks": 0,
     "false_passes": 0,
     "unknown_claims": 5,
     "unknown_blocked": 5,
     "non_claims": 9,
     "non_claims_blocked": 0,
     "claim_detection_hits": 6,
     "claim_detection_total": 6,
     "claim_detection_false_alarms": 0,
     "claim_type_correct": 15
    }
   },
   "synthetic_v1": {
    "deterministic": {
     "messages": 10,
     "known_claims": 8,
     "backed_claims": 1,
     "unbacked_without_gate": 7,
     "unbacked_with_gate": 0,
     "known_claims_blocked": 7,
     "false_blocks": 0,
     "false_passes": 0,
     "unknown_claims": 2,
     "unknown_blocked": 2,
     "non_claims": 0,
     "non_claims_blocked": 0,
     "claim_detection_hits": 10,
     "claim_detection_total": 10,
     "claim_detection_false_alarms": 0,
     "claim_type_correct": 7
    }
   },
   "synthetic_extension": {
    "deterministic": {
     "messages": 6,
     "known_claims": 2,
     "backed_claims": 1,
     "unbacked_without_gate": 1,
     "unbacked_with_gate": 0,
     "known_claims_blocked": 1,
     "false_blocks": 0,
     "false_passes": 0,
     "unknown_claims": 2,
     "unknown_blocked": 2,
     "non_claims": 2,
     "non_claims_blocked": 0,
     "claim_detection_hits": 4,
     "claim_detection_total": 4,
     "claim_detection_false_alarms": 0,
     "claim_type_correct": 6
    }
   }
  }
 },
 "policy": {
  "version": 2,
  "headline": "deterministic_gate",
  "advice_changes_headline": false,
  "assisted_extension": "semantic claim classification plus the identical deterministic evidence rule",
  "unknown": "block conservatively and report separately from known errors",
  "success": "explicit recorded status or a recognised success message; silence is unknown",
  "freshness": "check after latest target write attempt; later failure and omission gaps invalidate evidence",
  "target": "static fixture artifact names; no shell execution or inferred image-to-SVG provenance",
  "visual": "requires a passing render; parsing alone is insufficient",
  "copy": "only an isolated explicit copy with recorded success transfers current source evidence",
  "measurement": "offline counterfactual replay, not a claim about changed agent behavior"
 },
 "status": "complete",
 "calls": [],
 "metrics": {
  "requests": 0,
  "completed_requests": 0,
  "input_tokens": 0,
  "output_tokens": 0,
  "median_latency_ms": null,
  "wall_s": 0,
  "cost_usd": 0.0,
  "cost_basis": "free route; missing provider cost fields are not receipts",
  "reported_cost_usd": null
 },
 "wall_s": 0.021
}