https://kelet.ai/
Article
-
Kelet is an RCA agent that diagnoses why LLM app outputs fail
-
Uses ‘Prompt Patches’ as suggested fixes surfaced from root cause analysis
-
Targets silent failures — wrong answers, stuck loops — not just crashes
Discussion
-
Commenters ask how it handles non-prompt root causes like bad tool selection or retrieval ranking
-
Questions about supported frameworks and how generic the tool actually is
-
Differentiation from existing telemetry platforms (LangChain, Braintrust) unclear
Discuss on HN
