The autonomy level comes from observable signing patterns and anchored logs, not a self-declared label. The page also states plainly what the log cannot prove, such as actions taken off-log.
| Time (UTC) | Action | Amount | Signers | Human approved | Anchor |
|---|---|---|---|---|---|
| 14:02:11 | Swap USDC to SOL on Jupiter | 42.5 USDC | Agent key | No | Batch 19 |
| 13:48:40 | Add liquidity, SOL/USDC pool | 1.2 SOL + 180 USDC | Agent key + co-signer | Yes | Batch 19 |
| 13:15:03 | Swap SOL to USDC on Jupiter | 0.8 SOL | Agent key | No | Batch 19 |
| 12:31:57 | Withdraw to treasury | 3.0 SOL | Agent key + co-signer | Yes | Batch 18 |
| 11:58:22 | Rebalance to 60/40 target | 0.35 SOL | Agent key | No | Batch 18 |
| 11:02:09 | Transfer fee to ops wallet | 12 USDC | Agent key | No | Batch 18 |
Entries recorded by the SDK for all six agents, counted by d
Built for agent builders who want credibility, token communities checking autonomy claims, researchers and funds doing diligence on agent projects.
Call your existing agent function through log.run. Sundial signs a timestamped entry with the agent's logging key and queues it for the next anchor batch.
Planned
import { Sundial } from "@sundial/sdk";
const log = new Sundial({
agent: "9mWk3TqZ7bXvR2dNfJ5yHcA8sLePu4GxQ6tKwB1nDzYh",
signingKey: process.env.SUNDIAL_KEY!,
coSigner: "3vRcY8nKd2JpWq5XbF7mTzA4hGsE9uLx6NtQaH1yZeDk",
});
const sig = await log.run(
{ action: "swap", from: "USDC", to: "SOL", amount: 42.5 },
async () => executeSwap(route)
);
// entry is signed, timestamped and queued for batch anchoringEvery 15 minutes the SDK hashes the pending entries into a Merkle root and writes it in a Solana memo transaction. Editing an old entry changes its hash and breaks the match.
Planned| Batch | Entries | Merkle root | Memo tx | Status |
|---|---|---|---|---|
| 19 | 14 | a41f...c9e2 | 5Kq2...9vTn | Anchored |
| 18 | 11 | 7be0...31d8 | 2hXw...Rc4e | Anchored |
| 17 | 9 | d093...5ab7 | 4PmV...kT8s | Anchored |
| 16 | 12 | 18cc...e640 | 3JdN...Wq2y | Anchored |
| 20 | 3 | pending | none yet | Pending |
Anyone can open the agent page to see each action, the wallets and keys the agent controls, and whether a human approved that step.
PlannedThe badge is computed from the transactions themselves. If a human co-signer appears on the agent's transactions, the level drops. Nobody can pick their own level.
PlannedThe verifier fetches the wallet's transactions from Solana and compares them with the anchored log. It lists matches, unlogged transactions and edited entries.
PlannedThe autonomy level comes from observable signing patterns and anchored logs, not a self-declared label. The page also states plainly what the log cannot prove, such as actions taken off-log.
Every screen reads from the same model. Live parts read public onchain data; nothing ever asks your wallet to sign.
Read the docs{
"product": "Sundial",
"network": "solana",
"features": [
{
"name": "TypeScript SDK that wraps agent actions and emit",
"status": "planned"
},
{
"name": "Batched anchoring of log hashes in Solana memo t",
"status": "planned"
},
{
"name": "Public agent page listing the actions, the walle",
"status": "planned"
},
{
"name": "Embeddable autonomy badge (levels 0 to 3) comput",
"status": "planned"
}
]
}Live features work today. Planned ones are next on the roadmap, in this order.
Short answers. The docs have the rest.
A public, timestamped activity log and autonomy badge for onchain AI agents. The autonomy level comes from observable signing patterns and anchored logs, not a self-declared label. The page also states plainly what the log cannot prove, such as actions taken off-log.
Sundial is in development. Every feature is on the status list with what ships next, and the app opens in a sample workspace so you can see how it will work.
No. You can explore without one. Connecting a Solana wallet only shares your public address so Sundial can read your balances; it never asks you to sign a transaction.
Agent builders who want credibility, token communities checking autonomy claims, researchers and funds doing diligence on agent projects.
A public, timestamped activity log and autonomy badge for onchain AI agents.