Integrate with Codex
Use Codex to inspect and update your Agent project. Your running Agent Runtime, rather than the coding task, later receives Spendkit credentials and submits any authorized payment.
Put the project boundary in AGENTS.md
When this project needs AI Agent payments, read:
https://spendkit-alpha.vercel.app/docs/runtime/
https://spendkit-alpha.vercel.app/docs/examples/
https://spendkit-alpha.vercel.app/docs/api-quickstart/
https://spendkit-alpha.vercel.app/docs/mcp-tools/
Implement OAuth Client Credentials, MCP tool discovery, and a bound Payment Wallet adapter.
Start with read-only get_connection, get_policy, and preview_payment.
Keep real Client Secrets and Payment Wallet private keys out of source and model context.
Never create or increase token allowance or change Spending Policy without the wallet owner.
For payment, verify authorization.paymentWallet and transaction.from match the bound wallet,
submit the exact Router transaction, then record and reconcile its result.Review the integration in stages
- Ask Codex to identify the existing MCP client, tool loop, wallet sender, and secret storage.
- Have it implement a read-only connection and preview first. Inspect the Agent identity and bound wallet returned by Spendkit.
- Only after the owner completes dashboard wallet and Policy setup, enable the Runtime's wallet submission path and review the payment details before a testnet payment.
Remote MCP configuration is only one part.
A generic MCP connection does not automatically acquire and refresh Spendkit's OAuth Client Credentials token or submit a Router transaction from the bound wallet. Verify those capabilities in the Runtime you are building.