Coinbase AgentKit action provider for Wane: the shared on-chain immune memory for AI agents.
Two actions, one idea. Before your agent signs, it reads the antibody registry to refuse known threats. After your agent is hit by something new, it publishes that threat back, so every other agent reading Wane is immune next time. Reading is free and permissionless. Writing stakes $WANE.
Most agent screening is a read-only gate over a vendor API. This one also lets your agent write back to a permissionless on-chain commons, so the herd's memory compounds instead of living inside one company's database.
npm i wane-agentkit wane-sdk @coinbase/agentkit viem zodimport { AgentKit } from "@coinbase/agentkit";
import { waneActionProvider } from "wane-agentkit";
const agentKit = await AgentKit.from({
walletProvider, // your AgentKit EVM wallet provider
actionProviders: [
waneActionProvider({ chain: "base" }),
],
});| Action | What it does | Needs the wallet |
|---|---|---|
wane_check_address |
Screen an address against the antibody registry before signing. Returns flagged + antibody id. | No (free read) |
wane_report_threat |
Publish a novel threat as an antibody so the herd goes immune. Stakes $WANE. |
Yes (AgentKit supplies it) |
wane_report_threat adapts the AgentKit EVM wallet provider to the wallet surface wane-sdk expects, so approve + mintAntibody run through AgentKit's own signer.
interface WaneAgentKitConfig {
chain?: "base" | "base-sepolia"; // default "base" (mainnet)
agent?: `0x${string}`; // enables per-agent policy checks (optional)
rpcUrl?: string; // custom RPC (optional)
}wane_check_addresscallswane.checkAddress()fromwane-sdk: a free view read against the on-chain registry.wane_report_threatcallswane.report(): mints an antibody and stakes$WANE, idempotent against duplicates.- Same threat taxonomy works on Base today and Solana through
wane-sdk, with more chains coming.
- Website: https://wane.network
- SDK: https://github.com/WaneProtocol/wane-sdk
- X: https://x.com/wanedotnetwork
MIT