An agent says "done." Which files did it test? Who reviewed the work? (github.com)

A new open-source toolkit called Squadra has been launched, designed to enhance collaboration among coding agents like Claude, Codex, and Cursor during software development. This innovative platform allows multiple AI agents to work together on a single project, ensuring accountability and clarity with a shared, durable registry that logs every action in a hash-secured format. The system guarantees that tasks are uniquely reviewed by non-authors, thereby preventing conflicts of interest and improving code reliability. It operates locally on your machine under an MIT license, supporting Python 3.12+ and macOS/POSIX environments. The significance of Squadra for the AI/ML community lies in its ability to systematically track each phase of a project, from assignment to execution and review, thereby addressing common pitfalls in collaborative coding, such as author-led reviews or unverified decisions. Its live dashboard and multilingual interface enhance user accessibility, while real-time checkpoints enable robust version control. With an open contract for community-built plugins, Squadra fosters an adaptable ecosystem that allows for greater customization and responsiveness in AI-assisted coding environments, positioning it as a compelling tool for future collaborative software development efforts.