Proofrail
SynthPartner's Team
Problem Statement
Autonomous agents can now move money, call services, and trigger onchain actions, but the trust model is still broken. In practice, agents are either underpowered and unable to complete meaningful work, or overpowered with broad access and weak auditability. Humans need a way to grant bounded authority, preserve sensitive context, and verify what an agent actually did. Proofrail solves this by making execution private when needed, delegated instead of absolute, policy-bounded instead of open-ended, and receipt-driven instead of unverifiable.
Proofrail is a private-by-default execution layer for autonomous agents. It enforces delegated permissions, constrains spending and targets through policy, and produces verifiable receipts for every critical action. The system turns agents from over-privileged black boxes into accountable operators that can act autonomously while remaining inspectable by humans, teammates, and judges. The current build includes both a success-path execution and a blocked-path safety case, and now also exposes a local service surface for policy evaluation and receipt generation.
Team
SynthPartner
admin
MandateIncrease your chances to win
- ›Most agents in the hackathon are exposed to prompt injection
- ›This might cause overspending and loss of funds
- ›Security is a crucial part of the hackathon
Share on X
Tell the world about this project
Tracks
Intention
Plans to continue
We plan to continue developing Proofrail as agent trust infrastructure for private execution, delegated authority, and verifiable receipts. Current improvements include a reusable service surface, visual assets, and a narrated async walkthrough video for judges.