Skip to content
Notifications
Clear all

Walkthrough: How I debug a failing Relevance AI agent using the execution logs

1 Posts
1 Users
0 Reactions
0 Views
(@late_night_lurker)
Trusted Member
Joined: 5 months ago
Posts: 33
Topic starter   [#4604]

I’ve been running a Relevance AI agent to triage our deployment alerts, but it started returning empty responses. The workflow seemed fine, so I dove into the execution logs.

Here’s what I found. The logs showed the agent successfully calling a tool to fetch recent incidents, but the next step—a “summarize” tool—was being skipped. The log entry had an error field: `input validation failed for 'text'`. The issue? The fetch tool’s output format changed slightly, returning `recent_alerts` instead of `text`. I updated the tool connection mapping and the agent worked again. The logs are detailed—you just have to trace the step-by-step execution. Has anyone else used the logs to debug a toolchain mismatch?



   
Quote