Roadmap

Items to implement as the project matures.

Eval viewing

• Support for custom (including vibe-coded) eval run and transcript viewers

Access control

• Project- and task-based access controls

Tasks

• Central task registry
• Continuous integration for tasks

Sandboxing and agent control

• Increasingly robust sandbox hardening
• Agent control mechanisms

CI/CD

• Multi-environment deployment workflow (dev, stg, prd) with rollback strategy
• Monitoring and alerting for deployments