Skip to content

Admin Dashboard: Tool Call Log

Tool Call Log is the execution-level record of tools invoked by agents. Use it to debug an individual integration call; use Audit Log for administrative configuration changes.

Filter by agent name, tool name, session ID, status, and the From/To date range. Results show timestamp, agent, tool, status, duration, argument summary, and result summary. Select a row to inspect the complete stored values.

  1. Start with the session ID from the chat or task.
  2. Filter to failed calls and inspect the exact tool name and duration.
  3. Compare arguments with the tool’s discovered schema.
  4. Check MCP Servers for current connection/catalog state and Resilience for an open breaker.
  5. Re-run only after correcting the cause.

The expected result is a specific failure boundary: model-selected arguments, SyntheticBrew execution, transport/authentication, or upstream tool behavior.

Arguments and results may contain customer or business data. Limit Admin access and avoid copying entries into unsecured support channels. Transport errors returned to end users are sanitized, while operators should use server logs for the underlying network detail.