Buildkite MCP Server can now find your slowest and flakiest tests

The Buildkite MCP Server now includes list_tests, a read-only tool that gives AI agents a suite-wide view of your Test Engine tests and their execution metrics.

Agents can sort tests by reliability or average, total, minimum, or maximum duration to quickly identify flaky and slow tests. They can narrow the results by time window, branch, labels, owners, state, or tags, and paginate through large test suites.

Each result includes the test details alongside its reliability, duration metrics, execution count, and counts by result. This gives agents the context to investigate test health without fetching and combining individual test runs first.

list_tests is available in Buildkite MCP Server v1.18.0 and the Remote MCP server. It is part of the Test Engine toolset and requires the read_suites scope.