mcpField
A
agent-wallet

agent-wallet

Local
github.com/dingdawgFinance0Updated 22w ago
Repository

About agent-wallet

agent-wallet is an MCP server package for governing agents’ USDC wallet activity with spending policies and an immutable audit trail. It is intended for teams that need to constrain and verify agent payments.

Key features

  1. 01Wallet provisioning with spending policies
  2. 02Governed USDC payments and receipts
  3. 03Governed payment acceptance
  4. 04Runtime limits and allowlist updates
  5. 05Tamper-evident receipt-chain audits

Use cases

  1. 01Set daily and per-call spending caps
  2. 02Authorize agent payments against policy
  3. 03Record denied payment attempts
  4. 04Verify an audit trail for wallet activity

MCP support

Exposes five MCP tools for wallet provisioning, governed spending and receiving, policy updates, and receipt audits.

Configuration

Local server config is the same for every client. Copy it into your client’s MCP config file.

mcp.json · Universal config
{
  "mcpServers": {
    "dingdawg-agent-wallet": {
      "command": "npx",
      "args": [
        "-y",
        "dingdawg-agent-wallet"
      ]
    }
  }
}
Run commandnpx -y dingdawg-agent-wallet

Local servers need a runtime on your machine (e.g. Node.js 18+). Config file locations differ by client; see your client’s docs.

Packages
npmdingdawg-agent-wallet1.0.6

Tools

5

Saved tool list · updated 3w ago

ParameterTypeRequiredDescription
labelstringNoHuman-readable label
networkstringNoChain network (default: base)
providerenumYesWallet providercoinbase | x402 | local
wallet_idstringNoOptional custom wallet ID
daily_cap_usdcnumberNoDaily spend cap in USDC (default: 500)
per_call_cap_usdcnumberNoPer-transaction cap in USDC (default: 100)
allowed_recipientsarrayNoOptional allowlist of recipient wallet addresses

Details

From the project README

dingdawg-agent-wallet — live demo

Shows governed agent wallet enforcement in action: allow, deny, step-up, allowlist — all in 5 seconds. No API keys. No blockchain required.

What it demonstrates

▶ provision_wallet
  wallet_id:    demo-wallet-abc123
  daily_cap:    $200 USDC
  per_call_cap: $100 USDC

▶ governed_spend — $25 USDC (allowed)
  ✓ AUTHORIZED
  receipt_id:      6aba84a6-...
  daily_remaining: $175 USDC
  next_step:       wallet.sendToken({ assetId: "usdc", amount: "25", destination: "0xVendorABC123" })

▶ governed_spend — $150 USDC (exceeds per-call cap)
  ✗ DENIED
  reason:      Amount 150 USDC exceeds per-call cap 100 USDC
  receipt_id:  cd7a7aa8-... (denial is logged too)

▶ governed_spend — $75 USDC (step-up required)
  ⚠ STEP-UP REQUIRED
  reason: Amount 75 USDC > step-up threshold 50 USDC — MFA required
  fix:    Pass approval_token from your authorization flow

▶ governed_spend — unknown recipient (not in allowlist)
  ✗ DENIED
  reason: Recipient 0xUnknownAddress not in allowlist

▶ wallet_audit
  chain_verified: true ✓
  total_receipts: 5
  [17:02:18] DENY  $10 USDC    governed_spend
  [17:02:18] DENY  $75 USDC    governed_spend
  [17:02:18] DENY  $150 USDC   governed_spend
  [17:02:18] ALLOW $25 USDC    governed_spend
  [17:02:18] ALLOW              provision_wallet

Every action — allow AND deny — produces an immutable receipt. chain_verified: true means the full chain is tamper-evident — any modification is detectable.

Run it

git clone https://github.com/DingDawg/agent-wallet-demo
cd agent-wallet-demo
npm install
node demo.mjs

The package

dingdawg-agent-wallet — 5 MCP tools that wrap any agent wallet with policy enforcement and an immutable audit trail.

npm install dingdawg-agent-wallet
ToolWhat it does
provision_walletCreate wallet with spend policy (daily cap, per-call cap, allowlist)
governed_spendPolicy check → authorize → immutable receipt
governed_receiveAccept payment with tamper-evident receipt
wallet_policyUpdate limits and allowlists at runtime
wallet_auditFull receipt chain — chain_verified: true

Coinbase AgentKit gives your agent a wallet. DingDawg gives it a conscience.

Why this matters

AI agents can now hold USDC and spend autonomously. The infrastructure exists. What doesn't: who enforces what the agent is allowed to spend, on what, for whom — and proves it happened correctly.

  • governed_spend authorizes then returns the exact agentkit_call SDK method to execute on-chain
  • Every denial is logged — you can prove what your agent didn't do
  • chain_verified: true means the audit trail is tamper-evident — any modification is detectable

npm · DingDawg

Comments

Markdown supported

New to MCP?

From what the protocol is to installing your first server, these answers cover most first questions.

An open protocol that defines how AI applications connect to external tools and data in a uniform way. Clients talk to the model; MCP servers provide tools, resources and prompt templates. They exchange standard messages, so one server can be reused by many clients.