Let agents execute.
Keep keys in the vault.
Server-side EVM signing with AWS KMS, protocol-aware policies, and full audit logs.
Trusted by the ecosystem
AI agents need signatures, not private keys
AI Agent directly accesses private keys
Private keys never leave the HSM
Unrestricted transaction signing
Deny-by-default policy engine
Unknown calldata silently passes
Fail-closed: unknown calldata is always rejected
No audit trail
Structured JSON audit logs
Trust Boundary
Every signing request passes through decode, policy evaluation, and audit — before the HSM ever signs a byte.
Policy Engine
Define exactly what your agent can and cannot do. Default-deny ensures nothing slips through.
{
"version": "1.0",
"defaultAction": "deny",
"rules": [
{
"name": "Allow USDC approve",
"action": "allow",
"conditions": {
"contract": "0xa0b86991c6218b36c1d19d4a2e9eb0ce3606eb48",
"method": "approve",
"maxAmount": "1000000000000"
}
},
{
"name": "Allow Uniswap swap",
"action": "allow",
"conditions": {
"contract": "0xE592427A0AEce92De3Edee1F18E0157C05861564",
"method": "exactInputSingle",
"maxSlippage": "0.5"
}
}
]
}Policy format version
Deny everything not explicitly allowed
Allowlist rules with conditions
Developer Interfaces
Use Agentic Vault however you prefer — as a TypeScript library, CLI tool, MCP server, or OpenClaw plugin.
import {
createSigningProvider,
EvmSignerAdapter
} from '@agenticvault/agentic-vault';
const provider = createSigningProvider({
provider: 'aws-kms',
keyId: 'arn:aws:kms:us-east-1:123456789:key/your-key-id',
region: 'us-east-1',
});
const signer = new EvmSignerAdapter(provider);
const address = await signer.getAddress();
console.log('Vault address:', address);Supported Protocols
Built-in decoders for popular DeFi protocols. Every calldata byte is decoded and validated before signing.
ERC-20
Allowance cap, spender/token allowlist
Uniswap V3
Token pair, slippage, recipient validation
Aave V3
Asset allowlist, rate mode, amount limits
Already using OpenClaw?
Add secure signing in minutes. Agentic Vault integrates as an OpenClaw plugin with 7 safe tools and 2 policy-gated tools.
Safe Tools
- ✓
get_address - ✓
health_check - ✓
get_balance - ✓
get_token_balance - ✓
decode_calldata - ✓
dry_run - ✓
list_policies
Policy-Gated Tools
- ⚠
sign_transaction - ⚠
sign_typed_data
Getting Started
Three steps to secure your AI agent's transaction signing.
Install
Configure
Set up plugin config with keyId, region, and policyConfigPath
Secure
Create a deny-by-default policy.json