Product evidence

Regression Case Files

Confirmed open-source regressions, replayed with Early and verified against upstream evidence.

A Pydantic AI Regression Inside a 47-File Fix

A streaming handler that returned early could leave an agent run with an incomplete or empty response.

Production optimization could remove a valid CommonJS default export, causing consumers' default imports to resolve to undefined at runtime.

A valid form request could fail with HTTP 500 or return HTTP 200 with corrupted fields.

Every published case follows the same evidence path.

1
Select the release window
Start with a confirmed regression and compare the last known-good release with the affected release.
2
Run Early
Analyze the repository changes without providing the later issue or fix as the expected answer.
3
Reproduce the behavior
Execute a focused reproduction across the known-good, affected, and fixed versions.
4
Match the evidence
Compare Early's finding with the public issue, introducing change, and upstream fix.

Use a release your team already understands to evaluate what Early finds.