This document serves as a comprehensive guide to the prioritized objectives of the GuardRail project. It offers insight into the direction of the project, aiding contributors in understanding its trajectory. It also helps contributors determine whether their contributions align with the project's long-term goals.
While a feature may not be listed here, it doesn't imply automatic refusal of a patch (except for "frozen features" mentioned below). We welcome patches for new features and encourage innovation. However, please be aware that such patches may take longer to review.
| Feature | Description | Status | Developer (GitHub ID) |
|---|---|---|---|
| Mapping of GuardRail questions to AI adversarial threats. | Curate a list of AI adversarial threats and map them to relevant GuardRail questions. | 🚧 To Do | TBD |
| Mapping of AI adversarial threats to CVEs. | Curate a list of CVEs related to AI/ML and map the relevant ones to the found AI adversarial threats. | 🚧 To Do | TBD |
| Feature | Description | Status | Developer (GitHub ID) |
|---|---|---|---|
| Mapping of AI adversarial threats to AI adversarial robustness tools. | Curate a list of AI adversarial robustness testing tools and map them to the found AI adversarial threats and relevant GuardRail questions. | 🚧 To Do | TBD |
| Partially automate GuardRail questions. | For the AI adversarial threat-related questions, use adversarial robustness testing tools to test the AI/ML systems and to answer those questions. | 🚧 To Do | TBD |
Legend:
- ✅ Done: Completed feature.
- 🚧 To Do: Feature in progress.
Contributors are welcome to explore and contribute to other issues on the GuardRail repository: GuardRail GitHub Issues