Agent skills

A Claude Code, Cursor, or Codex skill that drives @gethookmyapp/cli end-to-end. Your AI does the wiring.

Install the skill

The skill ships via the standard Claude Code skill-install flow. It works in Cursor and Codex as well, following each client's skill-loading convention.

npx skills add hookmyapp/agent-skills --all

What the skill knows how to do

  • Authenticate the CLI: runs hookmyapp login, surfaces the browser step to the human.
  • Connect your own WhatsApp number through Meta embedded signup or start a sandbox session.
  • Pull env values with hookmyapp sandbox env --write .env or hookmyapp channels env <channel>.
  • Set a webhook URL via hookmyapp channels webhook set and verify with channels webhook show.
  • Send a test message via hookmyapp sandbox send to prove the loop is working.

Browser-required steps

Some steps require a human in the browser.

hookmyapp login and hookmyapp channels connect open OAuth popups. The skill surfaces a HUMAN ACTION REQUIRED prompt and waits for you to complete the tab before moving on.

Safety rules the skill enforces

  • Never pastes access tokens into chat output. Tokens stay in .env files.
  • Never runs hookmyapp channels webhook set against your own channel without explicit confirmation.
  • Never runs hookmyapp channels disable without explicit confirmation.
  • Never modifies .env files without telling you which keys will change.

Next steps