persona-harness

Adoption verdict: persona-harness is an extremely early-stage open-source experiment, evidenced only by a GitHub repository with 13 stars and 0 forks. There are no issues, discussions, documentation, or external reviews. The concept of preventing AI agents from prematurely claiming completion is intriguing, but without any practical validation or community feedback, the project is suitable only for technical exploration and not recommended for production use.

What it does and how it works: The tool acts as a gating hook inside OpenCode’s agent workflow. When an agent attempts to emit a “Done!” message, persona-harness checks whether real test evidence (e.g., a test report file) is present; if not, it blocks the completion. It is specifically designed for Java/Spring stacks and requires manual configuration within OpenCode. The barrier to entry is low: the code is open-source and free, but you need familiarity with OpenCode and Spring project structures, and you must define what constitutes acceptable test evidence. Note that it does not run tests itself – it only verifies that evidence exists, and actual testing must be handled separately.