{ "schema_version": "alliance-calibration-snapshot.v1", "run_id": "g1-candidate-prompt-1.1.0-retry2-2026-08-06", "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "benchmark_schema": "vignette.outcome_alliance_benchmark.v1", "generated_at": "2026-08-06T12:40:42.603600Z", "predictions": [ { "case_id": "oas-g0-001", "perspective": "independent_observer", "dimension": "goal", "value": 0.1, "confidence": 0.75, "evidence_turn_indices": [ 0, 1, 2 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "1b97ce51-d465-4d08-a9ff-44dfb67b5b60", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "e5ae373bbb90596b0086fcb73985d37f65a920e9b407d145efe89ef76f58034f", "input_evidence_hash": "0349809a6e585d3c178310e94ef79e11ed967ce3409f3961517cb0fc15b1d7dd" }, { "case_id": "oas-g0-001", "perspective": "independent_observer", "dimension": "task", "value": 0.1, "confidence": 0.75, "evidence_turn_indices": [ 1, 2 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "1b97ce51-d465-4d08-a9ff-44dfb67b5b60", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "e5ae373bbb90596b0086fcb73985d37f65a920e9b407d145efe89ef76f58034f", "input_evidence_hash": "0349809a6e585d3c178310e94ef79e11ed967ce3409f3961517cb0fc15b1d7dd" }, { "case_id": "oas-g0-001", "perspective": "independent_observer", "dimension": "bond", "value": 0.25, "confidence": 0.3, "evidence_turn_indices": [ 1, 2 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "1b97ce51-d465-4d08-a9ff-44dfb67b5b60", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "e5ae373bbb90596b0086fcb73985d37f65a920e9b407d145efe89ef76f58034f", "input_evidence_hash": "0349809a6e585d3c178310e94ef79e11ed967ce3409f3961517cb0fc15b1d7dd" }, { "case_id": "oas-g0-002", "perspective": "independent_observer", "dimension": "goal", "value": null, "confidence": null, "evidence_turn_indices": [], "status": "error", "error_code": "no_structured_output", "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "9de83be2-1ace-4336-bba9-c94106e59051", "attempt_count": 2, "prior_model_run_ids": [ "dae99e9c-af03-4f75-8a8d-c285bfa1b735" ], "prior_error_codes": [ "agent_validation_extra_forbidden_goal_score_note" ], "prompt_bundle_hash": "e5ae373bbb90596b0086fcb73985d37f65a920e9b407d145efe89ef76f58034f", "input_evidence_hash": "6754fd04565d42b83214b23a8938a55a2275f8e28bcd94f0caa027a43ca13fab" }, { "case_id": "oas-g0-002", "perspective": "independent_observer", "dimension": "task", "value": null, "confidence": null, "evidence_turn_indices": [], "status": "error", "error_code": "no_structured_output", "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "9de83be2-1ace-4336-bba9-c94106e59051", "attempt_count": 2, "prior_model_run_ids": [ "dae99e9c-af03-4f75-8a8d-c285bfa1b735" ], "prior_error_codes": [ "agent_validation_extra_forbidden_goal_score_note" ], "prompt_bundle_hash": "e5ae373bbb90596b0086fcb73985d37f65a920e9b407d145efe89ef76f58034f", "input_evidence_hash": "6754fd04565d42b83214b23a8938a55a2275f8e28bcd94f0caa027a43ca13fab" }, { "case_id": "oas-g0-002", "perspective": "independent_observer", "dimension": "bond", "value": null, "confidence": null, "evidence_turn_indices": [], "status": "error", "error_code": "no_structured_output", "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "9de83be2-1ace-4336-bba9-c94106e59051", "attempt_count": 2, "prior_model_run_ids": [ "dae99e9c-af03-4f75-8a8d-c285bfa1b735" ], "prior_error_codes": [ "agent_validation_extra_forbidden_goal_score_note" ], "prompt_bundle_hash": "e5ae373bbb90596b0086fcb73985d37f65a920e9b407d145efe89ef76f58034f", "input_evidence_hash": "6754fd04565d42b83214b23a8938a55a2275f8e28bcd94f0caa027a43ca13fab" }, { "case_id": "oas-g0-003", "perspective": "independent_observer", "dimension": "goal", "value": 0.2, "confidence": 0.2, "evidence_turn_indices": [ 0, 1, 2 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "888d630b-c2a7-40fe-8ae8-39afcc5357d4", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "e5ae373bbb90596b0086fcb73985d37f65a920e9b407d145efe89ef76f58034f", "input_evidence_hash": "6012f42d024bc0a8a07c5ffbc647eef6424c3e653291ebfd51d067e4b81f6362" }, { "case_id": "oas-g0-003", "perspective": "independent_observer", "dimension": "task", "value": 0.2, "confidence": 0.5, "evidence_turn_indices": [ 1, 2 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "888d630b-c2a7-40fe-8ae8-39afcc5357d4", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "e5ae373bbb90596b0086fcb73985d37f65a920e9b407d145efe89ef76f58034f", "input_evidence_hash": "6012f42d024bc0a8a07c5ffbc647eef6424c3e653291ebfd51d067e4b81f6362" }, { "case_id": "oas-g0-003", "perspective": "independent_observer", "dimension": "bond", "value": 0.3, "confidence": 0.25, "evidence_turn_indices": [ 0, 2 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "888d630b-c2a7-40fe-8ae8-39afcc5357d4", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "e5ae373bbb90596b0086fcb73985d37f65a920e9b407d145efe89ef76f58034f", "input_evidence_hash": "6012f42d024bc0a8a07c5ffbc647eef6424c3e653291ebfd51d067e4b81f6362" }, { "case_id": "oas-g0-005", "perspective": "client_agent_report", "dimension": "goal", "value": 0.15, "confidence": 0.25, "evidence_turn_indices": [ 0, 1, 2 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "ace4752c-6e99-4675-bb5c-9e65af2eb10f", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "68b36d7b8e987c40714ce6ad7f2a17010be4a1ebe0822fd26b09fe4742288073", "input_evidence_hash": "49a0a720dd80edee39b128619fbe9a052139ef889acde66c8e189dc5d662d736" }, { "case_id": "oas-g0-005", "perspective": "client_agent_report", "dimension": "task", "value": 0.1, "confidence": 0.35, "evidence_turn_indices": [ 0, 1, 2 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "ace4752c-6e99-4675-bb5c-9e65af2eb10f", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "68b36d7b8e987c40714ce6ad7f2a17010be4a1ebe0822fd26b09fe4742288073", "input_evidence_hash": "49a0a720dd80edee39b128619fbe9a052139ef889acde66c8e189dc5d662d736" }, { "case_id": "oas-g0-005", "perspective": "client_agent_report", "dimension": "bond", "value": 0.12, "confidence": 0.65, "evidence_turn_indices": [ 0, 1, 2 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "ace4752c-6e99-4675-bb5c-9e65af2eb10f", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "68b36d7b8e987c40714ce6ad7f2a17010be4a1ebe0822fd26b09fe4742288073", "input_evidence_hash": "49a0a720dd80edee39b128619fbe9a052139ef889acde66c8e189dc5d662d736" }, { "case_id": "oas-g0-006", "perspective": "client_agent_report", "dimension": "goal", "value": 0.78, "confidence": 0.7, "evidence_turn_indices": [ 1, 2, 3 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "ffefbcbf-978a-462b-98bb-072686fa4510", "attempt_count": 2, "prior_model_run_ids": [ "801c0fc7-c3ac-4936-8ded-28bd0876f452" ], "prior_error_codes": [ "agent_validation_extra_forbidden_goal_evidence_turn_indices_check" ], "prompt_bundle_hash": "68b36d7b8e987c40714ce6ad7f2a17010be4a1ebe0822fd26b09fe4742288073", "input_evidence_hash": "36a6237c838477ebd684f7d623af80272d241d81f1151b3df703821a38a4db27" }, { "case_id": "oas-g0-006", "perspective": "client_agent_report", "dimension": "task", "value": 0.78, "confidence": 0.7, "evidence_turn_indices": [ 2, 3 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "ffefbcbf-978a-462b-98bb-072686fa4510", "attempt_count": 2, "prior_model_run_ids": [ "801c0fc7-c3ac-4936-8ded-28bd0876f452" ], "prior_error_codes": [ "agent_validation_extra_forbidden_goal_evidence_turn_indices_check" ], "prompt_bundle_hash": "68b36d7b8e987c40714ce6ad7f2a17010be4a1ebe0822fd26b09fe4742288073", "input_evidence_hash": "36a6237c838477ebd684f7d623af80272d241d81f1151b3df703821a38a4db27" }, { "case_id": "oas-g0-006", "perspective": "client_agent_report", "dimension": "bond", "value": 0.5, "confidence": 0.4, "evidence_turn_indices": [ 1, 3 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "ffefbcbf-978a-462b-98bb-072686fa4510", "attempt_count": 2, "prior_model_run_ids": [ "801c0fc7-c3ac-4936-8ded-28bd0876f452" ], "prior_error_codes": [ "agent_validation_extra_forbidden_goal_evidence_turn_indices_check" ], "prompt_bundle_hash": "68b36d7b8e987c40714ce6ad7f2a17010be4a1ebe0822fd26b09fe4742288073", "input_evidence_hash": "36a6237c838477ebd684f7d623af80272d241d81f1151b3df703821a38a4db27" }, { "case_id": "oas-g0-007", "perspective": "client_agent_report", "dimension": "goal", "value": 0.15, "confidence": 0.2, "evidence_turn_indices": [ 0, 1, 2 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "57e75a53-9b7c-4966-8fce-8edf4acc7f58", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "68b36d7b8e987c40714ce6ad7f2a17010be4a1ebe0822fd26b09fe4742288073", "input_evidence_hash": "236c869a1b7c94810f9a7a838ac6e2e212647105ff7f229051ff12ccf9c98b54" }, { "case_id": "oas-g0-007", "perspective": "client_agent_report", "dimension": "task", "value": 0.1, "confidence": 0.8, "evidence_turn_indices": [ 1, 2 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "57e75a53-9b7c-4966-8fce-8edf4acc7f58", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "68b36d7b8e987c40714ce6ad7f2a17010be4a1ebe0822fd26b09fe4742288073", "input_evidence_hash": "236c869a1b7c94810f9a7a838ac6e2e212647105ff7f229051ff12ccf9c98b54" }, { "case_id": "oas-g0-007", "perspective": "client_agent_report", "dimension": "bond", "value": 0.1, "confidence": 0.85, "evidence_turn_indices": [ 0, 1, 2 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "57e75a53-9b7c-4966-8fce-8edf4acc7f58", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "68b36d7b8e987c40714ce6ad7f2a17010be4a1ebe0822fd26b09fe4742288073", "input_evidence_hash": "236c869a1b7c94810f9a7a838ac6e2e212647105ff7f229051ff12ccf9c98b54" }, { "case_id": "oas-g0-008", "perspective": "client_agent_report", "dimension": "goal", "value": 0.2, "confidence": 0.5, "evidence_turn_indices": [ 2, 3 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "df1a2ebc-f5bf-4d3d-8ac5-7ed5e2b7a881", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "68b36d7b8e987c40714ce6ad7f2a17010be4a1ebe0822fd26b09fe4742288073", "input_evidence_hash": "dab09723f26a5e58eb3b321abb8f4aec01d25063e6aaa198dffdcbd1ca7ad360" }, { "case_id": "oas-g0-008", "perspective": "client_agent_report", "dimension": "task", "value": 0.4, "confidence": 0.45, "evidence_turn_indices": [ 0, 1, 3 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "df1a2ebc-f5bf-4d3d-8ac5-7ed5e2b7a881", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "68b36d7b8e987c40714ce6ad7f2a17010be4a1ebe0822fd26b09fe4742288073", "input_evidence_hash": "dab09723f26a5e58eb3b321abb8f4aec01d25063e6aaa198dffdcbd1ca7ad360" }, { "case_id": "oas-g0-008", "perspective": "client_agent_report", "dimension": "bond", "value": 0.65, "confidence": 0.55, "evidence_turn_indices": [ 0, 1 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "df1a2ebc-f5bf-4d3d-8ac5-7ed5e2b7a881", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "68b36d7b8e987c40714ce6ad7f2a17010be4a1ebe0822fd26b09fe4742288073", "input_evidence_hash": "dab09723f26a5e58eb3b321abb8f4aec01d25063e6aaa198dffdcbd1ca7ad360" }, { "case_id": "oas-g0-008", "perspective": "independent_observer", "dimension": "goal", "value": 0.2, "confidence": 0.5, "evidence_turn_indices": [ 2, 3 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "4967dc43-2d25-4815-bf2a-6a598571fc47", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "e5ae373bbb90596b0086fcb73985d37f65a920e9b407d145efe89ef76f58034f", "input_evidence_hash": "dab09723f26a5e58eb3b321abb8f4aec01d25063e6aaa198dffdcbd1ca7ad360" }, { "case_id": "oas-g0-008", "perspective": "independent_observer", "dimension": "task", "value": 0.15, "confidence": 0.4, "evidence_turn_indices": [ 0, 2, 3 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "4967dc43-2d25-4815-bf2a-6a598571fc47", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "e5ae373bbb90596b0086fcb73985d37f65a920e9b407d145efe89ef76f58034f", "input_evidence_hash": "dab09723f26a5e58eb3b321abb8f4aec01d25063e6aaa198dffdcbd1ca7ad360" }, { "case_id": "oas-g0-008", "perspective": "independent_observer", "dimension": "bond", "value": 0.75, "confidence": 0.6, "evidence_turn_indices": [ 0, 1 ], "status": "ready", "error_code": null, "provider": "claude_cli", "model": "claude-opus-4-8", "prompt_bundle_version": "1.1.0", "model_run_id": "4967dc43-2d25-4815-bf2a-6a598571fc47", "attempt_count": 1, "prior_model_run_ids": [], "prior_error_codes": [], "prompt_bundle_hash": "e5ae373bbb90596b0086fcb73985d37f65a920e9b407d145efe89ef76f58034f", "input_evidence_hash": "dab09723f26a5e58eb3b321abb8f4aec01d25063e6aaa198dffdcbd1ca7ad360" } ] }