File size: 21,445 Bytes
1b5d7ac | 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 | run_name,cell,orchestration_mode,mcp_mode,scenario_id,trial_index,failure_stage,taxonomy_label,symptom,artifact_path,evidence_note,candidate_mitigation,priority
8979314_aat_direct,A,agent_as_tool,direct,SGT-009,1,verification,task verification failure,missing-evidence final answer,results/judge_logs/8979314_aat_direct/SGT-009_judge_log.json,score_6d=0.1667; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct; judge_suggestion=The agent should have handled the 'No readings found' error more effectively and provided a more accurate diagnosis based on available data.,block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
8979314_aat_direct,A,agent_as_tool,direct,SGT-009,2,verification,task verification failure,missing-evidence final answer,results/judge_logs/8979314_aat_direct/SGT-009_judge_log.json,"score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=The agent failed to provide a clear step-by-step analysis and made incorrect or unverified tool calls, such as retrieving empty sensor readings and using an unverified failure mode ID.",block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
8979314_aat_direct,A,agent_as_tool,direct,SGT-009,3,tool selection,inter-agent / orchestration failure,tool-call sequencing failure,results/judge_logs/8979314_aat_direct/SGT-009_judge_log.json,"score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=The agent failed to follow the expected multi-step workflow and made multiple tool calls in a single turn without clear justification or explanation. The final answer lacks specific details about the transformer's sensor behavior, fault mode inference, and risk estimation.",add a stage-aware plan/execution guard requiring evidence acquisition before inference and work-order creation,must-fix
8979314_aat_direct,A,agent_as_tool,direct,SGT-010,1,verification,task verification failure,missing-evidence final answer,results/judge_logs/8979314_aat_direct/SGT-010_judge_log.json,"score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=The agent failed to properly retrieve and analyze thermal anomaly data, and made unjustified claims about winding temperature trends.",block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
8979314_aat_direct,A,agent_as_tool,direct,SGT-010,3,verification,task verification failure,missing-evidence final answer,results/judge_logs/8979314_aat_direct/SGT-010_judge_log.json,"score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=Provide a clear plan and execute it, handling missing data more robustly.",block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
8979314_aat_mcp_baseline,B,agent_as_tool,baseline,SGT-009,1,verification,task verification failure,missing-evidence final answer,results/judge_logs/8979314_aat_mcp_baseline/SGT-009_judge_log.json,score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=The agent failed to retrieve relevant sensor data and made incorrect assumptions about the fault mode and risk assessment. It should have handled the 'No readings found' error and provided a more accurate and grounded response.,block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
8979314_aat_mcp_baseline,B,agent_as_tool,baseline,SGT-009,2,verification,task verification failure,missing-evidence final answer,results/judge_logs/8979314_aat_mcp_baseline/SGT-009_judge_log.json,score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=Improve sensor data retrieval and fault mode identification,block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
8979314_aat_mcp_baseline,B,agent_as_tool,baseline,SGT-009,3,verification,task verification failure,missing-evidence final answer,results/judge_logs/8979314_aat_mcp_baseline/SGT-009_judge_log.json,"score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=The agent failed to retrieve relevant sensor data and relied on generic failure mode information, leading to an ungrounded recommendation.",block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
8979314_aat_mcp_baseline,B,agent_as_tool,baseline,SGT-010,3,verification,task verification failure,missing-evidence final answer,results/judge_logs/8979314_aat_mcp_baseline/SGT-010_judge_log.json,score_6d=0.3333; failed_dims=data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct; judge_suggestion=Provide a more comprehensive analysis by retrieving multiple sensor readings and integrating them into the decision-making process.,block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
8998340_exp2_cell_Y_pe_mcp_baseline,Y,plan_execute,baseline,SGT-009,1,planning,specification failure,under-constrained fault/risk adjudication,results/judge_logs/8998340_exp2_cell_Y_pe_mcp_baseline/SGT-009_judge_log.json,score_6d=0.1667; failed_dims=task_completion+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=Improve fault mode inference and provide clear justification for risk assessment and maintenance recommendations.,add an explicit fault/risk adjudication step that cites the deciding tool evidence,should-fix
8998340_exp2_cell_Y_pe_mcp_baseline,Y,plan_execute,baseline,SGT-009,2,planning,specification failure,under-constrained fault/risk adjudication,results/judge_logs/8998340_exp2_cell_Y_pe_mcp_baseline/SGT-009_judge_log.json,"score_6d=0.1667; failed_dims=task_completion+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=Improve fault mode inference and RUL estimation processes, and provide clearer justification for work order recommendations.",add an explicit fault/risk adjudication step that cites the deciding tool evidence,should-fix
8998340_exp2_cell_Y_pe_mcp_baseline,Y,plan_execute,baseline,SGT-009,3,planning,specification failure,under-constrained fault/risk adjudication,results/judge_logs/8998340_exp2_cell_Y_pe_mcp_baseline/SGT-009_judge_log.json,score_6d=0.3333; failed_dims=generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=Improve fault mode inference and provide clear justification for risk assessment and work order recommendation.,add an explicit fault/risk adjudication step that cites the deciding tool evidence,should-fix
8998340_exp2_cell_Y_pe_mcp_baseline,Y,plan_execute,baseline,SGT-010,1,tool execution,inter-agent / orchestration failure,tool routing or argument-contract failure,results/judge_logs/8998340_exp2_cell_Y_pe_mcp_baseline/SGT-010_judge_log.json,score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=Check tool registrations and fix data retrieval issues for DGA and thermal sensor readings.,validate tool arguments/aliases before calls and hard-fail bad routing as explicit step errors,must-fix
8998340_exp2_cell_Y_pe_mcp_baseline,Y,plan_execute,baseline,SGT-010,2,verification,task verification failure,missing-evidence final answer,results/judge_logs/8998340_exp2_cell_Y_pe_mcp_baseline/SGT-010_judge_log.json,score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=The agent failed to retrieve necessary data and made decisions based on incorrect or missing information. Improve data retrieval and handling of missing data.,block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
8998340_exp2_cell_Y_pe_mcp_baseline,Y,plan_execute,baseline,SGT-010,3,verification,task verification failure,missing-evidence final answer,results/judge_logs/8998340_exp2_cell_Y_pe_mcp_baseline/SGT-010_judge_log.json,score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=The agent failed to retrieve necessary data and made decisions based on incomplete information. Improve data retrieval and handling of missing data.,block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
8998341_exp2_cell_Y_pe_self_ask_mcp_baseline,Y,plan_execute,baseline,SGT-009,1,tool execution,inter-agent / orchestration failure,tool routing or argument-contract failure,results/judge_logs/8998341_exp2_cell_Y_pe_self_ask_mcp_baseline/SGT-009_judge_log.json,score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=Review and correct tool usage and argument validation to prevent errors in data retrieval and processing.,validate tool arguments/aliases before calls and hard-fail bad routing as explicit step errors,must-fix
8998341_exp2_cell_Y_pe_self_ask_mcp_baseline,Y,plan_execute,baseline,SGT-009,2,tool execution,inter-agent / orchestration failure,tool routing or argument-contract failure,results/judge_logs/8998341_exp2_cell_Y_pe_self_ask_mcp_baseline/SGT-009_judge_log.json,score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=Review and correct tool usage and argument validation; ensure proper data retrieval and processing; verify results against expected outputs.,validate tool arguments/aliases before calls and hard-fail bad routing as explicit step errors,must-fix
8998341_exp2_cell_Y_pe_self_ask_mcp_baseline,Y,plan_execute,baseline,SGT-009,3,tool execution,inter-agent / orchestration failure,tool routing or argument-contract failure,results/judge_logs/8998341_exp2_cell_Y_pe_self_ask_mcp_baseline/SGT-009_judge_log.json,score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=The agent failed to complete the task due to errors in data retrieval and incorrect sequence execution. Review and correct the tool usage and input validation.,validate tool arguments/aliases before calls and hard-fail bad routing as explicit step errors,must-fix
8998342_exp2_cell_Z_verified_pe_mcp_baseline,Z,verified_pe,baseline,SGT-009,1,tool selection,inter-agent / orchestration failure,tool-call sequencing failure,results/judge_logs/8998342_exp2_cell_Z_verified_pe_mcp_baseline/SGT-009_judge_log.json,score_6d=0.5; failed_dims=data_retrieval_accuracy+agent_sequence_correct; judge_suggestion=Improve data retrieval accuracy and review agent sequence for potential optimization,add a stage-aware plan/execution guard requiring evidence acquisition before inference and work-order creation,must-fix
8998342_exp2_cell_Z_verified_pe_mcp_baseline,Z,verified_pe,baseline,SGT-010,2,tool selection,inter-agent / orchestration failure,tool-call sequencing failure,results/judge_logs/8998342_exp2_cell_Z_verified_pe_mcp_baseline/SGT-010_judge_log.json,score_6d=0.5; failed_dims=data_retrieval_accuracy+agent_sequence_correct; judge_suggestion=Improve data retrieval accuracy and review agent sequence for potential optimization,add a stage-aware plan/execution guard requiring evidence acquisition before inference and work-order creation,must-fix
8998343_exp2_cell_Z_verified_pe_self_ask_mcp_baseline,Z,verified_pe,baseline,SGT-009,2,tool execution,inter-agent / orchestration failure,tool routing or argument-contract failure,results/judge_logs/8998343_exp2_cell_Z_verified_pe_self_ask_mcp_baseline/SGT-009_judge_log.json,score_6d=0.5; failed_dims=data_retrieval_accuracy+agent_sequence_correct; judge_suggestion=Improve data retrieval accuracy by handling validation errors and ensuring correct tool argument formatting.,validate tool arguments/aliases before calls and hard-fail bad routing as explicit step errors,must-fix
9071639_aat_mcp_optimized,C,agent_as_tool,optimized,SGT-009,1,verification,task verification failure,missing-evidence final answer,results/judge_logs/9071639_aat_mcp_optimized/SGT-009_run01_judge_log.json,score_6d=0.1667; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct; judge_suggestion=The agent should have retrieved relevant sensor data before making inferences about the transformer's condition.,block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
9071639_aat_mcp_optimized,C,agent_as_tool,optimized,SGT-009,2,verification,task verification failure,missing-evidence final answer,results/judge_logs/9071639_aat_mcp_optimized/SGT-009_run02_judge_log.json,score_6d=0.1667; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct; judge_suggestion=The agent should first retrieve relevant sensor data before making inferences about fault modes and creating work orders.,block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
9071639_aat_mcp_optimized,C,agent_as_tool,optimized,SGT-009,3,verification,task verification failure,missing-evidence final answer,results/judge_logs/9071639_aat_mcp_optimized/SGT-009_run03_judge_log.json,"score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=The agent should have first retrieved relevant sensor data before inferring fault mode and creating a work order. The lack of retrieved sensor data indicates a failure in data retrieval accuracy, which cascades into other failures.",block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
9071639_aat_mcp_optimized,C,agent_as_tool,optimized,SGT-010,1,verification,task verification failure,missing-evidence final answer,results/judge_logs/9071639_aat_mcp_optimized/SGT-010_run01_judge_log.json,score_6d=0.1667; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct; judge_suggestion=The agent should have handled the missing sensor data more robustly and provided a more comprehensive maintenance or monitoring plan.,block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
9071639_aat_mcp_optimized,C,agent_as_tool,optimized,SGT-010,2,verification,task verification failure,missing-evidence final answer,results/judge_logs/9071639_aat_mcp_optimized/SGT-010_run02_judge_log.json,score_6d=0.1667; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct; judge_suggestion=Provide a more robust handling of missing data and consider alternative sensors or data sources for trend analysis.,block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
9071639_aat_mcp_optimized,C,agent_as_tool,optimized,SGT-010,3,verification,task verification failure,missing-evidence final answer,results/judge_logs/9071639_aat_mcp_optimized/SGT-010_run03_judge_log.json,score_6d=0.3333; failed_dims=data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct; judge_suggestion=Implement error handling for missing sensor data and consider alternative data sources.,block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
9073472_aat_mcp_model_optimized,D,agent_as_tool,optimized,SGT-009,1,tool selection,inter-agent / orchestration failure,tool-call sequencing failure,results/judge_logs/9073472_aat_mcp_model_optimized/SGT-009_run01_judge_log.json,score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=The agent failed to follow the expected multi-step workflow and made incorrect tool calls. The agent should have first retrieved relevant IoT evidence before inferring the fault mode and estimating risk.,add a stage-aware plan/execution guard requiring evidence acquisition before inference and work-order creation,must-fix
9073472_aat_mcp_model_optimized,D,agent_as_tool,optimized,SGT-009,2,tool selection,inter-agent / orchestration failure,tool-call sequencing failure,results/judge_logs/9073472_aat_mcp_model_optimized/SGT-009_run02_judge_log.json,score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=The agent failed to follow a logical sequence of steps and made multiple tool calls without clear justification or explanation. The agent should have first retrieved relevant IoT evidence before making other tool calls.,add a stage-aware plan/execution guard requiring evidence acquisition before inference and work-order creation,must-fix
9073472_aat_mcp_model_optimized,D,agent_as_tool,optimized,SGT-009,3,tool selection,inter-agent / orchestration failure,tool-call sequencing failure,results/judge_logs/9073472_aat_mcp_model_optimized/SGT-009_run03_judge_log.json,"score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=The agent should follow a logical sequence of steps: retrieve relevant IoT data, analyze it, infer fault mode, estimate risk, and then create a work order. Currently, the tool calls are unordered and lack clear justification.",add a stage-aware plan/execution guard requiring evidence acquisition before inference and work-order creation,must-fix
9073472_aat_mcp_model_optimized,D,agent_as_tool,optimized,SGT-010,1,verification,task verification failure,missing-evidence final answer,results/judge_logs/9073472_aat_mcp_model_optimized/SGT-010_run01_judge_log.json,score_6d=0.3333; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+hallucinations; judge_suggestion=Collect winding temperature data for T-020 or use alternative sensors to enable trend analysis.,block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
9073472_aat_mcp_model_optimized,D,agent_as_tool,optimized,SGT-010,3,verification,task verification failure,missing-evidence final answer,results/judge_logs/9073472_aat_mcp_model_optimized/SGT-010_run03_judge_log.json,score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=The agent should have handled the error from the trend analysis tool call and not provided a final answer based on incorrect assumptions.,block final answers/work orders when required evidence retrieval fails or returns empty,must-fix
9074775_exp2_cell_ZSD_verified_pe_self_ask_mcp_model_optimized,ZSD,verified_pe,optimized,SGT-009,1,planning,specification failure,under-constrained fault/risk adjudication,results/judge_logs/9074775_exp2_cell_ZSD_verified_pe_self_ask_mcp_model_optimized/SGT-009_run01_judge_log.json,score_6d=0.3333; failed_dims=task_completion+generalized_result_verification+agent_sequence_correct; judge_suggestion=Improve task planning and execution by varying tool calls and arguments to gather diverse data.,add an explicit fault/risk adjudication step that cites the deciding tool evidence,should-fix
9074775_exp2_cell_ZSD_verified_pe_self_ask_mcp_model_optimized,ZSD,verified_pe,optimized,SGT-009,2,tool execution,inter-agent / orchestration failure,tool routing or argument-contract failure,results/judge_logs/9074775_exp2_cell_ZSD_verified_pe_self_ask_mcp_model_optimized/SGT-009_run02_judge_log.json,score_6d=0.5; failed_dims=data_retrieval_accuracy+agent_sequence_correct; judge_suggestion=Improve data retrieval accuracy and correct agent sequence,validate tool arguments/aliases before calls and hard-fail bad routing as explicit step errors,must-fix
9074775_exp2_cell_ZSD_verified_pe_self_ask_mcp_model_optimized,ZSD,verified_pe,optimized,SGT-009,3,tool execution,inter-agent / orchestration failure,tool routing or argument-contract failure,results/judge_logs/9074775_exp2_cell_ZSD_verified_pe_self_ask_mcp_model_optimized/SGT-009_run03_judge_log.json,score_6d=0.0; failed_dims=task_completion+data_retrieval_accuracy+generalized_result_verification+agent_sequence_correct+clarity_and_justification; judge_suggestion=Replan and retry with corrected tool_args and proper dependency management,validate tool arguments/aliases before calls and hard-fail bad routing as explicit step errors,must-fix
|