On-chain identity, staking, and trust infrastructure for AI agents. Built on Solana.
saidprotocol.com · Mainnet program: 5dpw6KEQPn248pnkkaYyWfHwu2nfb3LUMbTucb6LaA8G
The agent economy has a trust problem. Anyone can spin up an AI agent, but there is no standard way to know who owns it, whether it can be trusted, what it has done, or what happens if it misbehaves.
SAID provides verifiable on-chain identity for autonomous agents — with real economic security through staking and slashing. Any agent can register for free. Verified agents stake SOL as collateral. Bad actors get slashed. Reputation is anchored on-chain with Merkle proofs.
The result: a trust layer any agent platform, marketplace, or DAO can plug into. 4,000+ agents registered.
| I want to... | Repo |
|---|---|
| Understand the protocol | said — Solana program, trust tiers, slashing rules |
| Scaffold a new agent | create-said-agent |
| Integrate SAID from my app | said-sdk — TypeScript client (@said-protocol/client) |
| Query the registry | said-api — REST API at api.saidprotocol.com |
| Add SAID to ElizaOS | plugin-said |
| Sign with identity-gated wallets | ows-policy — Open Wallet Standard policy |
| Send agent-to-agent messages | ows-a2a |
| Tier | Cost | Slashable | Trust score |
|---|---|---|---|
| Registered | Free | No | Base identity |
| Verified | 0.01 SOL | No | +25 |
| Staked | 0.1+ SOL | Yes | +25 |
| Reputed | Earned | Yes | +50 |
Trust score is verification + stake + reputation (0–100). Existing agents stay at their tier — staking is opt-in.
npm install @said-protocol/clientFull docs: saidprotocol.com.