Add validated Krea2 eval recorder

tools/krea2_record_eval.py

@@ -0,0 +1,46 @@
+#!/usr/bin/env python3
+from __future__ import annotations
+
+import argparse
+import json
+import sys
+from pathlib import Path
+
+
+ROOT = Path(__file__).resolve().parents[1]
+if str(ROOT) in sys.path:
+    sys.path.remove(str(ROOT))
+sys.path.insert(0, str(ROOT))
+
+import krea2_eval_log  # noqa: E402
+
+
+def _load_entry(path: Path) -> dict:
+    with path.open("r", encoding="utf-8") as handle:
+        data = json.load(handle)
+    if not isinstance(data, dict):
+        raise ValueError("entry JSON must contain one object")
+    return data
+
+
+def main() -> int:
+    parser = argparse.ArgumentParser(description="Validate and append one durable Krea2 fixed-seed eval entry.")
+    parser.add_argument("--entry-json", required=True, help="Path to a JSON object containing one eval entry.")
+    parser.add_argument("--log-path", default=str(krea2_eval_log.DEFAULT_EVAL_LOG_PATH), help="Eval log path to update.")
+    parser.add_argument("--dry-run", action="store_true", help="Validate without writing the log.")
+    args = parser.parse_args()
+
+    try:
+        entry = _load_entry(Path(args.entry_json))
+        log = krea2_eval_log.append_entry(entry, path=args.log_path, dry_run=args.dry_run)
+    except Exception as exc:
+        print(f"error: {exc}", file=sys.stderr)
+        return 1
+
+    action = "validated" if args.dry_run else "recorded"
+    print(f"{action}: {entry.get('id')} ({len(log.get('entries') or [])} entries)")
+    return 0
+
+
+if __name__ == "__main__":
+    raise SystemExit(main())

tools/prompt_smoke.py

@@ -6994,6 +6994,49 @@ def smoke_krea2_eval_log_policy() -> None:
     mutation["observation"] = "mutation should not leak"
     clean = krea2_eval_log.entries_for_variant("pov_handjob_upright_centered")[0]
     _expect(clean.get("observation") != "mutation should not leak", "Krea2 eval log leaked caller mutation")
+    with tempfile.TemporaryDirectory() as tmpdir:
+        temp_log_path = Path(tmpdir) / "krea2-eval-log.json"
+        temp_log_path.write_text(json.dumps(log, ensure_ascii=True, indent=2) + "\n", encoding="utf-8")
+        smoke_entry = {
+            "id": "ballsucking-9001-low-head-smoke",
+            "date": "2026-06-29",
+            "variant_key": "pov_ballsucking_low_head",
+            "seed": 9001,
+            "source": "smoke",
+            "result": "inconclusive",
+            "decision": "needs_more_tests",
+            "baseline_prompt_summary": "Baseline prompt kept the head too high for the atlas low-head target.",
+            "candidate_prompt_summary": "Candidate prompt moved the head below the shaft at testicle height.",
+            "observation": "Smoke entry validates the durable fixed-seed record path without changing the real eval log.",
+            "baseline_image": "/tmp/krea2_baseline.png",
+            "candidate_image": "",
+            "commit": "smoke",
+        }
+        errors = krea2_eval_log.validate_entry(
+            smoke_entry,
+            existing_entries=log.get("entries") or [],
+            catalog_keys=set(krea2_pose_variant_catalog.variant_keys()),
+        )
+        _expect(errors == [], f"Valid Krea2 eval entry failed validation: {errors}")
+        appended_log = krea2_eval_log.append_entry(smoke_entry, path=temp_log_path)
+        _expect(len(appended_log.get("entries") or []) == len(entries) + 1, "Krea2 eval append did not add one entry")
+        appended_entries = krea2_eval_log.entries_for_variant("pov_ballsucking_low_head", path=temp_log_path)
+        _expect(appended_entries and appended_entries[-1].get("seed") == 9001, "Krea2 eval append did not persist temp entry")
+        duplicate_errors = krea2_eval_log.validate_entry(
+            smoke_entry,
+            existing_entries=appended_log.get("entries") or [],
+            catalog_keys=set(krea2_pose_variant_catalog.variant_keys()),
+        )
+        _expect(any("duplicate id" in error for error in duplicate_errors), "Krea2 eval validation should reject duplicate ids")
+        bad_variant = dict(smoke_entry)
+        bad_variant["id"] = "missing-variant-9001"
+        bad_variant["variant_key"] = "missing_variant"
+        bad_variant_errors = krea2_eval_log.validate_entry(
+            bad_variant,
+            existing_entries=appended_log.get("entries") or [],
+            catalog_keys=set(krea2_pose_variant_catalog.variant_keys()),
+        )
+        _expect(any("unknown variant" in error for error in bad_variant_errors), "Krea2 eval validation should reject unknown variants")
 
 
 def smoke_krea2_prompt_guide_policy() -> None: