Code Repositories

A connected code repository unlocks code correlation in your analyses. Without it, the diagnosis can only reason about symptoms in your logs. With it, the diagnosis becomes specific — turning "timeout in checkout flow" into "the regression is in PR #482, which changed how expired auth tokens are handled in services/billing/charge.ts."

Supported platforms

After a repo is connected, configure Service mappings so the AI knows which service in your logs corresponds to which path inside which repo.

What a connected repo enables

When a repo is connected, your analyses can include code correlation — identifying the file, function, or pull request likely behind a symptom in your logs.

Choosing an auth method

Each platform supports a recommended path and a fallback. The recommended path is always more secure and supports better revocation; the fallback is simpler if you can't go through the recommended path for organizational reasons.

Lookback window

Each repo connection has a lookback days setting (default 30, configurable 1–365). This is how far back analyses will look when correlating symptoms to recent merges. For high-deploy-cadence teams, a shorter window is fine. For repos that change infrequently, push it longer to give analyses more history to draw from.

Quotas

The number of repos per project depends on your plan:

If you have more services than your plan allows, group multiple services into a single connected repo using service mappings.

Force-refresh repo data

Repo data refreshes automatically over time. If you need to force a refresh — after a force-push, credential rotation, or large refactor — use Force refresh under each repo's settings in the dashboard.