Microsoft has open sourced code-testing-generator, a polyglot agent that writes unit tests and proves they work, shipping in the dotnet-test plugin inside the MIT-licensed dotnet/skills repository. The agent addresses the gap where prompts like ‘generate unit tests’ do not specify framework, file location, or assertions; it reads the repository before writing, then plans, writes, runs, and checks tests. On Microsoft’s internal 152-task benchmark, it completed 140 tasks against 120 for stock GitHub Copilot, with both setups using the same model and prompts. It is an agent definition with skills, not a hosted service, so it runs inside existing coding agents with code staying local. It is viable from solo maintainers upward, with startups and mid-market teams gaining most; enterprises can fork language guidance. Industries include regulated or audit-heavy estates like financial services, healthcare, insurance, and public sector, plus platform teams. Applications include backfilling tests, generating tests for pull-request diffs, raising coverage before release gates, and standardising conventions. The agent coordinates through a Research-Plan-Implement pipeline, detecting language and test framework, reading existing tests, and finding real build and test commands. It picks from three strategies: Direct, Single pass, or Iterative, never modifying production code and avoiding tests with external URLs, bound ports, or timing dependencies. Before reporting completion, it runs five checks, including reasoning about small code changes that should fail tests