Skip to content

Tool execution logs

Execution logs record every tool run triggered by assistants — parameters sent, HTTP status, duration, errors, and transform outcomes. Use them to debug production issues without exposing secrets to visitors.

  1. Go to Tools → select a tool → Logs.
  2. Filter by time range, status (success/failure), or deployment.
  3. Open a row for request summary, response size, and error detail.

Org-wide activity may also appear in the dashboard audit feed.

FieldDescription
TimestampWhen the run started
Assistant / deploymentWhich surface triggered the call
ParametersPath, query, body values (secrets redacted)
HTTP statusUpstream response code
DurationRound-trip latency
ErrorValidation failure, timeout, or API message
Rich contentWhether transform produced cards
PatternLikely cause
Spike in 401Expired OAuth or rotated API key
Strict validation errorsModel supplying wrong types — tune descriptions
200 but no cardsitems_path mismatch
TimeoutsSlow API — async mode + notifications

Log retention depends on your plan. Export via Tools API for longer archival if needed.

Credentials and raw auth headers are never stored in logs. Parameter values may be truncated per data handling settings.