WhatsApp API Skill: build WhatsApp integrations with your AI agent

You describe what you want on WhatsApp — an integration, an automation, or a small service — and your AI agent knows how to get there on 1MSG instead of guessing API calls in the chat. The 1msg-integration skill is for that: fewer dead ends on webhooks, delivery, and retries, so you move faster whether you ship code every day or build through vibe-coding.

What the skill does

It focuses on practices that are easy to skip on a first build: choosing the right tool path, verifying sends and delivery, handling webhooks, and recovering when something fails. Request syntax stays in the 1MSG API reference; the skill keeps the agent on the paths that work for 1MSG.

Add the skill to an AI workflow in Cursor, Codex, Claude Code, or another host that supports agent skills. The skill runs inside the agent; you stay in your editor or chat and steer the build.

Pair it with 1MSG clients in the order that matches how you build:

• MCP — the agent calls the API directly from the IDE or chat.
• CLI — the same operations from scripts, with less token use than MCP; use it when the job is fixed and you already know which methods you will run.
• SDK — typed client in your repo when you are shipping a long-lived integration, not a one-off agent session.

How to start

New project. For a new project, the skill leads you through channel credentials and API access (MCP, CLI, or SDK), then a verified outbound message and an inbound reply through your webhook — the minimum loop the skill treats as a working integration before you rely on it in production.

Existing codebase. You can point the skill at a project that already uses 1MSG. It helps you check the setup against 1MSG practices, spot common failure modes, and debug sends, webhooks, and delivery when something stops working.

Getting started: github.com/1msg/1msg-for-agents