๐Ÿ”ด AgentTriage AMD Developer Cloud

Multi-agent SRE incident response pipeline โ€” Planner โ†’ Executor โ†’ Summarizer

๐Ÿ” Groq API Key

๐ŸŽฎ Run the Agent

Single Crash
๐ŸŸข Easy ยท 20% noise
Cascading Failure
๐ŸŸก Medium ยท 30% noise
Silent Degradation
๐Ÿ”ด Hard ยท 60% noise

๐Ÿ“ก Pipeline Output

Select a task and click Run
Waiting for agent run...

๐Ÿ“Š Scores

Single Crash
โ€”
Cascading Failure
โ€”
Silent Degradation
โ€”

๐Ÿ—๏ธ Architecture

PLANNER โ†’ reads logs โ†’ decides severity + root cause + strategy
    โ†“
EXECUTOR โ†’ step loop โ†’ classify โ†’ identify โ†’ remediate โ†’ resolve
    โ†“
SUMMARIZER โ†’ generates structured incident report
    โ†“
Episode Score (0.0 โ†’ 1.0)
AMD MI300X Qwen2.5-72B LangGraph FastAPI Docker vLLM Python 3.11 OpenEnv