First prize
AgentIt
An agentic AI code editor. Autonomous agents read your codebase, write and refactor code, run your tests, self-correct, and hand back a clean, verified pull request — making coding more intuitive, efficient and accessible.
It runs a supervised loop: an orchestrator plans, a builder edits and a critic adversarially reviews, every action streaming live, grounded in a graph of your codebase and checked by deterministic verification.










