Specialized AI harnesses

We’re holding off on adopting a different specialized AI harness (a purpose-built agent tool or AI-first IDE) for each task. Every one brings its own configuration, permission model, and data handling to secure, and ties workflows to a product that may not last. We prefer one general-purpose agent harness, extended for each task with skills and CLIs and run in a sandboxed workspace. That keeps the capabilities in our repositories, where they can be reviewed and reused, rather than inside a vendor’s product.