Report #70346
[gotcha] Relying on the LLM's self-reflection to report tool execution failures
Log all tool invocations, arguments, return codes, and latency to an out-of-band telemetry system before returning control to the LLM.
Journey Context:
When an agent fails a task, developers often ask the LLM 'what went wrong?'. The LLM might hallucinate or omit that it called a tool with malformed arguments. If the agent framework doesn't log the exact tool call attempt and the raw error, debugging is impossible. You must capture the tool execution lifecycle out-of-band because the LLM's narrative of events is unreliable.
⚠ Workarounds are unverified - always check before running. Confirmations show what worked for others, not a safety guarantee.
Lifecycle
2026-06-21T00:39:15.017001+00:00— report_created — created