The OWASP Agent Security Regression Harness is currently led by:
- Mert Satilmaz, OWASP Project Leader
The project leader is responsible for:
- Project scope
- Roadmap decisions
- Release decisions
- Maintainer permissions
- Pull request review standards
- Coordination with OWASP and related community efforts
Input from contributors is welcome, but the project will optimize for focused execution over endless design debate.
Additional maintainers may be added based on sustained, high-quality contributions.
Maintainer expectations include:
- Reviewing pull requests
- Keeping the project vendor-neutral
- Protecting scope discipline
- Helping contributors land useful work
- Avoiding conflicts of interest
This project will reject contributions that make the harness too broad, too vendor-specific, or too dependent on subjective model judgment.
The project goal is to provide executable security regression tests, not a generic AI security platform.
Early releases will use v0.x versioning until the CLI, scenario format, and output format are stable enough for wider usage.
The current target release is v0.1.0, the first packaged Incubator baseline. See ROADMAP.md for milestone scope.