Skip to main content

One-prompt setup

Install the agent skills, then paste this into Claude Code, Cursor, or Codex. The skill tells the agent which commands to run and what to ask you before touching live settings.
Five ways to give an agent HookMyApp. Use as many as fit your setup: the CLI and Project MCP operate your account, the docs sources give the agent accurate context.

1. CLI

For anything with a terminal. The CLI owns credentials, tunnels, and webhook configuration.

2. Agent skills

Install once, and every future session knows the HookMyApp workflows, commands, and safety rules.

3. Project MCP

Live tools for agents with no shell: send messages, read delivery logs, manage channels and webhooks, publish to Instagram.
hookmyapp agent setup configures this in every coding agent installed on your machine: Claude Code, Codex, and Cursor. Add --client claude|codex|cursor to set up only one. The credential is written into each entry, so no client needs a separate sign-in; restart the client to pick the server up. For anything else, add the URL and sign in from the client, or use an hmok_ API key.

4. Docs MCP

Documentation search for agents, with no access to your account.
Read-only and scoped to published documentation. It also carries a submit_feedback tool, so an agent that hits a page that is wrong or unclear can tell us.

5. llms.txt

The whole documentation set as one file, for agents that take plain context.
llms.txt is the page index; llms-full.txt is the full text. Point an agent at either when it needs product context rather than live account access.

Which to use

Next steps