Investigate bugs with Oncall

Oncall investigates a bug and keeps the findings in a report. Start here when the cause is uncertain or the team needs reproduction evidence before choosing a fix.

Supply the incident contextCopy Anchor Link

Open the Oncall workspace and start a conversation. Include the observed behavior, expected behavior, affected repositories, approximate occurrence time, and reproduction steps. Attach a sanitized screenshot, clip, or log excerpt when useful.

Investigate the activity list sometimes staying empty after a filter reset.
Expected: clearing the query restores results within the current category.
The attached clip reproduces it. Find the cause and record evidence before changing code.

Separate observations from hypotheses. “The response was empty” is evidence; “the cache is broken” is a theory until the investigation supports it.

Read the reportCopy Anchor Link

Open the linked bug report and review the reproduction, findings, cause, supporting evidence, and proposed delivery. Follow related sessions for commands and checks. Record a missing environment or failed reproduction as a limitation.

If an investigation needs a production tool, configure the appropriate external tool connection and credentials. Connecting a repository alone does not provide production logs or service access.

Continue to a fixCopy Anchor Link

Use the report's linked work or ask for a bounded implementation once the cause is understood. Follow the resulting task and session, then review the change against the reproduction and acceptance evidence.

For automatic intake from new grouped issues, see Sentry. That connection is read-only investigation access; it does not grant unrestricted mutation of Sentry.