"gamAI Trust Layer team on Vultr: Agent Rush Hackathon Hackathon"
Team Idea
Evidence-first runtime for AI agents: every "done" is a verifiable claim. Fail-closed evidence gate, prompt-injection refusal memory, and a reproducible false-success benchmark. Sandboxed execution on Vultr (Serverless Inference + Cloud Compute), MIT-licensed, SHA-256 evidence per action.
Team Members
.png&w=256&q=75)
Renato Suster
renatogama
Submission
Submission draft in progress