Initialize AutoGen Hardware Postmortem Team

Linked challenge: AutoGen Hardware Kernel Postmortem Analyzer & Fault Injector

Prompt source

implementation14 lines4 sectionsNo variables0 checklist items

import autogen

config_list = [{'model': 'gpt-4o', 'api_key': 'ENV_KEY'}]

# Diagnostic Agent specializing in CUDA/Triton log parsing
kernel_expert = autogen.AssistantAgent(
    name="KernelExpert",
    system_message="You analyze CUDA stack traces, Triton PTX dumps, and NVML logs to identify GPU kernel failure causes.",
    llm_config={'config_list': config_list}
)

# System Recovery Agent
ops_engineer = autogen.AssistantAgent(
    name="OpsEngineer",
    system_message="You generate precise remediation steps (draining node, patching kernel args, dynamic fallback).",
    llm_config={'config_list': config_list}
)