AI Security Working Group

Attributes

Context

The introduction of the Duo Agent Platform moves from separate GitLab Duo product features to a dedicated platform for AI at GitLab.

Secure development of GitLab Duo features becomes more critical for the business as we launch the Duo Agent Platform. We can expect to see rapid iteration of this platform and should leverage industry best practices and make features secure by default.

This working group will include a community of team members from Engineering and Security whom strive to make security simple for users and contributors alike.

Scope

The scope of this group includes the following GitLab components:

1. AI Gateway
2. CLI agents
3. Duo Agent Platform
4. Editor Extensions
5. GitLab Language Server

Exit Criteria

1. Best practices for implementing AI prompts are documented in our Contributor documentation.
2. Proof-of-concepts are executed and recorded to understand what AI security tooling could offer SaaS, Dedicated, and Self-Managed customers.
3. Our CI/CD pipeline will trigger code review of merge requests and provide actionable advice for contributors.
4. Our CI/CD pipeline will block merge requests that do not meet secure development standards we establish for our AI offerings.
5. Automated scripts are established to setup local working environment and help in testing AI features on the various AI projects.

Roles and Responsibilities

View page source - Edit this page - please contribute.