{"record": {"aliases": ["MirrorCode"], "artifacts": [{"kind": "paper", "url": "https://arxiv.org/abs/2606.30182"}], "audits": [{"audit_date": "2026-07-17", "audit_url": "https://github.com/kimjune01/mirrorcode-audit", "auditor": "June Kim", "finding": "June Kim's audit reports that MirrorCode's claimed construct is autonomous reconstruction of whole programs from behavior, while its metric scores scoped reimplementation from a working oracle, partly via recall: the benchmark's own memorization screen flags 17 of 25 targets, and 2 of 25 are recall by construction (brotlid, mailauth). The audit discloses that Kim applied for a role at Epoch AI, which co-produced MirrorCode.", "receipts": ["https://github.com/kimjune01/mirrorcode-audit"]}], "categories": ["coding_agent"], "description": {"en": "25 target programs graded by exact output match against the reference on end-to-end tests, including held-out tests. Attempts are long-horizon and budget-sensitive: the paper reports $2,600 over 19 days for a single attempt on a large task, so two reported scores are comparable only when the inference budgets behind them are similar."}, "documents": [], "evidence_summary": {"document_count": null, "model_count": null, "model_count_basis": null}, "key": "model-reports:mirrorcode", "modality": null, "name": "MirrorCode", "openness": {"code_license": null, "data_license": null, "evidence": [], "status": "unknown"}, "provenance": {"registry": "data/model_cards.yml", "score_archive": "data/benchmark_scores.yml", "source_url": "https://arxiv.org/abs/2606.30182"}, "publisher": null, "released": "2026-06-29", "released_reference": {"basis": "benchmark_release", "source_key": "model-reports:mirrorcode", "source_url": "https://arxiv.org/abs/2606.30182"}, "schema_version": 1, "sizes": [], "slug": "mirrorcode", "source": "model_reports", "source_benchmark_id": "mirrorcode"}, "schema_version": 1, "scores_by_source": {}, "siblings": [{"key": "claire-radar:2606.30182", "name": "MirrorCode", "relation": "equivalent", "slug": "claire-radar-2606-30182", "source": "claire_radar"}]}