GitHub has launched an experimental Rubber Duck mode within the GitHub Copilot CLI. The newest addition to the AI-powered coding software makes use of a second mannequin from a distinct AI household to supply a second opinion earlier than enacting the agent’s plan.
The brand new characteristic was introduced April 6. Launched in experimental mode, Rubber Duck leverages a second mannequin from a distinct AI household to behave as an impartial reviewer, assessing plans and work on the moments the place suggestions issues most, in accordance with GitHub. Rubber Duck is a centered evaluate agent, powered by a mannequin from a complementary household to a main Copilot session. The job of Rubber Duck is to verify the agent’s work and current a brief, centered checklist of high-value issues together with particulars the first agent could have missed, assumptions price questioning, and edge instances to contemplate.
Builders can use/experimentalwithin the Copilot CLI to entry Rubber Duck alongside different experimental options.
