Help Center / Understand and Improve Agents
TroubleshootingRead an Agent Trace
Use traces, the step-by-step run logs, to find tool calls, failures, and ways to improve an agent.
On This PageOpen the Trace
Open the Trace
From an agent conversation or execution, open the trace view. A trace is the step-by-step run log; its timeline shows the request, reasoning milestones, tool calls, and final output.
Scan for missing inputs, failed tool calls, or places where the agent guessed instead of asking a clarifying question.
Improve the Agent
Update the agent instructions when the trace reveals a repeatable gap. Prefer one clear rule over a long rewrite.
Run the same task again so you can compare the before and after behavior.
Was this helpful?
Your feedback helps us rank and improve Help Center articles.
Related Articles
Meet Your Super Agent Command Center
Use your Super Agent to coordinate work, monitor progress, and launch repeatable tasks from one place.
Read ArticleCreate Your First Agent With Your Super Agent
Ask your Super Agent to create a focused agent for one team role.
Read ArticleGive Your Agents Long-Term Memory
Add, check, update, and remove agent memories so agents can reuse stable context safely.
Read ArticleCreate and Reuse Prompt Templates
Save proven requests as templates so teammates can run the same work without rewriting prompts.
Read ArticleSchedule Recurring Tasks and Automations
Turn a proven task into a scheduled job with dry runs, triggers, time zones, and run history.
Read ArticleDeploy an Always-On Hosted Super Agent
Keep a tested Super Agent online for shared team work, tools, and owners.
Read Article