Flaky Tests: Investigation limits and clearer analysis status

Automatic flaky test investigations now have monthly per-repository and per-test limits, and the Analysis tab explains why an analysis was skipped or failed.

Automatic flaky test investigations are now rate-limited per calendar month, so a burst of newly flaky tests can’t consume an unbounded amount of analysis.

Two limits apply, and both reset on the first of the month:

Limit	Default
Investigations per repository, per month	100
Investigations per individual test, per month	5

What this means for you:

• Automatic and MCP-triggered investigations count. Investigations Trunk starts when a test moves from Healthy to Flaky or Broken, and ones started through the fix-flaky-test MCP tool, count against both limits. Only investigations that finish and produce a result are counted.
• Manual investigations are always available. Investigations you start from a test’s detail page in the Trunk app are exempt from both limits, even after a repository or test hits its cap.
• Clearer status when an analysis doesn’t run. The Analysis tab now shows a banner explaining why an analysis was skipped, such as the monthly limit being reached or a previous run still in progress.
• Limits can be raised. If you expect higher volume, such as driving fixes with an agent through MCP, contact [e-mail].

See the Autofix Flaky Tests docs for full details.