Agents will run
the next economy.
Autonomous agents are already trading, paying, and negotiating on Solana. This isn't a future prediction — it's happening now. But without an identity layer, every agent is a black box. No one knows who's behind it, what it's allowed to do, or whether it's earned any trust.
We believe agents deserve real identity — and humans deserve real privacy. Not iris scans. Not social graphs. One zero-knowledge proof that says "a human is behind this" without revealing who. That's the foundation for a trustworthy agent ecosystem.
A ZK proof generated in your browser. No data collected, no identity revealed. Just a cryptographic commitment that a real person is behind this wallet.
Give your agent exactly the permissions it needs — and nothing more. Which programs, how much SOL, when it expires. You stay in control.
Any protocol on Solana can check if an agent is human-backed in one call. A new primitive for trust that the entire ecosystem can build on.
Four programs.
One stack.
Calls flow top → bottom. No circular dependencies. External programs only touch L4.The only layer external programs touch. CPI call returns is_valid, delegation scope, reputation score, and expiry timestamp. Reads from L3, L2, L1.
Score 0—10,000 computed from on-chain behavior reports. Any program can submit reports. Stake-based dispute resolution. Age bonus. Public and queryable.
Scoped permission boundaries for AI agents. Whitelist programs, cap SOL, set expiry, restrict actions. Max 5 concurrent delegations per Writ. Revocable.
ZK Groth16 proof verification on BN254. Poseidon nullifier storage prevents double-registration. SBT minting via Token-2022 NonTransferable. The foundation.
Zero-Knowledge Identity
Prove you're human without revealing who. Groth16 proofs generated in your browser, verified on-chain. Your identity never touches Solana. Only a Poseidon nullifier — one hash — lives on-chain.
Scoped Delegation
Your agent only does what you allow. Define exact boundaries: which programs it can call, how much it can spend, when access expires. Not a blank check — a programmable permission system.
Reputation Engine
Trust is earned, not declared. On-chain behavior scoring from 0 to 10,000. Any program can report, any program can query. Dispute resolution through staking. The longer you're here, the more weight you carry.
One CPI Integration
The entire integration surface is one function call. Any Solana program can verify an agent's identity, scope, and reputation in a single cross-program invocation. No middleware. No off-chain dependency.
No Biometrics
We don't want your face, your iris, or your fingerprint. One Poseidon hash prevents double-registration without linking to any real-world identity. The most private identity system possible.
Solana Native
Not ported from EVM. Built from scratch for Solana's architecture and compute constraints. Anchor 0.30, Token-2022 soulbound tokens, ark-bn254 pairings — all within 200k compute units.
One CPI call.
Full verification.
Any Solana program can verify an agent in a single cross-program invocation. Returns identity status, delegation scope, reputation score, and expiry. No SDK. No oracle. No off-chain dependency.
Every component is open source. No proprietary dependencies. No off-chain services. The entire protocol runs on Solana.
The agent era
starts with trust.
We're building the identity primitive that the agent ecosystem is missing. Open source. On-chain. Privacy-first. If you believe agents deserve accountability without surveillance — build with us.