What Happens When You Give AI Agents the Map of Your Code’s Coverage?
This title could be clearer and more informative.Try out Clickbait Shieldfor free (5 uses left this month).
JetBrains is introducing a new 'finding-tests' agent skill in Rider 2026.2 EAP that uses dotCover coverage data to help AI agents locate the correct test file when generating unit tests for .NET projects. Instead of having the agent scan the entire codebase to find where tests belong, Rider runs a coverage analysis and provides the exact file path, reducing token consumption by up to 50% in internal benchmarks. The skill ships as a bundled feature in Rider's AI Assistant and as a standalone MCP tool for external agents like Claude Code or Codex. A key trade-off is time: dotCover must run a full coverage scan, which can take seconds to hours depending on project size. Known issues include first-run failures, Codex bundled skill limitations, and 120-second timeouts for large solutions. Future plans include a 'target coverage' feature where AI agents automatically generate enough tests to hit a specified coverage percentage.
Table of contents
Why AI agents need more than just code contextWhat are agent skills?Enter the finding-tests skillThe result: Token costs are halvedTime vs. tokensKnown issuesWhat’s next on the roadmapTell us what you thinkSort: