Use Privy CLI with Claude Code, Codex or Cursor

The CLI is built so a coding agent can use it without guessing: it outputs JSON, explains every command through privy cli search and privy schema, previews changes before making them and reads them back afterwards.

Install the skill

npx skills add awesamarth/privy-cli

The skill tells Claude Code, Codex, Cursor and other agents when to reach for the CLI, how to check that you’re logged in, and to start from privy cli search. It’s short and readable.

Without the skill, name the tool in your request so the agent knows to look for it:

Use the Privy CLI to create an app called my-new-test and enable Google login.

What you still do yourself

• Log in. Run privy auth login in your own terminal. It needs a browser CAPTCHA and an email code, and the agent should never ask you for the code in chat. After that, the agent can renew the session on its own.
• Decide where secrets go. When an app or API key is created, you choose whether the secret goes to an env file, your OS keychain, both or neither.
• Approve anything big. Deleting apps, account-wide changes and production apps need extra confirmation flags, and the agent should only use them when you’ve asked for that change.

What the agent does

1. Finds the command: privy cli search "enable google login"
2. Reads the contract: privy schema apps config update
3. Previews: --dry-run offline, or --diff against your live app
4. Runs it with --yes
5. Reports what the CLI verified, and anything it couldn’t

If a change fails or times out, the CLI never retries on its own, so the agent checks the current state before trying again.

What it won’t do

The CLI manages your Privy developer account and apps. It doesn’t sign transactions, move funds, export private keys or touch billing. For an agent that needs its own wallet, Privy has a separate, official agent wallet CLI.