Zavu Functions

Zavu Functions let you run TypeScript in our cloud and wire it directly to your WhatsApp/SMS/Email senders. You write the code locally, run npx zavudev deploy, and seconds later your agent is live with tools backed by real business logic.

import { defineAgent, defineTool } from "@zavudev/functions"

defineAgent({
  senderId: process.env.SENDER_ID!,
  name: "Bella",
  provider: "zavu",
  model: "openai/gpt-4o-mini",
  prompt: "Eres Bella, anfitriona de la Pizzeria. Sé breve.",
  channels: ["whatsapp"],
})

defineTool({
  name: "check_availability",
  description: "Get free reservation slots for a date and party size.",
  parameters: {
    type: "object",
    properties: {
      date: { type: "string" },
      partySize: { type: "number" },
    },
    required: ["date", "partySize"],
  },
  handler: async ({ date, partySize }) => {
    // your real booking logic — DB, POS, external API, whatever
    return { available: true, slots: ["19:00", "21:00"] }
  },
})

That’s the entire integration. One file. npx zavudev deploy. Done.

When to use Functions

AI agent with real actions

Customer asks something, the LLM decides which tool to call, your code runs, real data goes back as the answer.

Custom webhooks

Receive inbound from Stripe, GitHub, your own systems, then forward as a WhatsApp message.

Scheduled jobs

Daily report digests, reminder broadcasts, cleanup tasks.

Event-driven automation

React to message.inbound, broadcast.status_changed, etc. with custom logic without managing webhook receivers yourself.

How it relates to AI Agents

Zavu has two complementary ways to build agents:

You want	Use
A no-code agent configured from the dashboard	AI Agents
An agent with custom business logic in your own code	Zavu Functions (this guide)
Both — start no-code, evolve to code	Start with AI Agents, migrate when you need tools or custom flows

Functions can fully manage an AI Agent: the defineAgent call creates and keeps the agent config in sync with your code. Once your function declares defineAgent, the agent is “managed by the function” — the dashboard surfaces this and disables manual edits to prevent drift.

Mental model

+----------------+        +--------------------+        +-----------+
| WhatsApp user  | -----> | Zavu sender (WABA) | -----> | AI Agent  |
+----------------+        +--------------------+        +-----------+
                                                              |
                                                       (tool call)
                                                              v
                                            +-------------------------------+
                                            | Your Zavu Function            |
                                            |  defineTool(handler)          |
                                            +-------------------------------+
                                                              |
                                                       (return result)
                                                              v
                                                       Agent answers user

Fully managed runtime. Each function runs in Zavu’s serverless cloud. No consoles to log into — npx zavudev deploy handles bundling, dependencies, and publishing.
Internal invocation. When the agent calls a tool, we use signed internal invocations — your function is not publicly exposed. No HTTP, no HMAC secrets to rotate, no DDoS surface.
Native event binding. Functions can also subscribe to Zavu events (message.inbound, broadcast.status_changed, etc) via triggers.
Auto-provisioned credentials. Every function gets a scoped ZAVUDEV_API_KEY in its environment so it can call our SDK without you handling key distribution.

Quick tour

Want your AI coding agent (Claude Code, Cursor, Copilot…) to write the function for you? Install Zavu’s Coding Agent Skills — the functions skill teaches it defineAgent, defineTool, npx zavudev deploy, and the full debug flow.

npx skills add zavudev/zavu-skills

Install the CLI

npx zavudev@latest --version
# or install it globally: npm install -g zavudev
# or grab a standalone binary from
# https://github.com/zavudev/zavu-cli/releases

npx zavudev login

Opens your browser, lets you pick a project. The API key is saved to ~/.zavu/credentials.json and used by every later command.

Scaffold a function

npx zavudev fn init --template restaurant-booking -y
cd reservations

Creates index.ts with defineAgent + defineTool boilerplate.

Set secrets

npx zavudev fn secrets set SENDER_ID jn76vnxet8g5nq661by3v06y1581bmmn

The agent will run on this WhatsApp sender. Get the ID from npx zavudev senders list.

Deploy

npx zavudev deploy

Bundle, upload, agent + tools synced to your sender.

Try it

Send a WhatsApp to the sender’s number. The agent answers, calls your tool handlers, returns real data.

Continue

Quickstart

Full step-by-step from zero to live agent in 10 minutes.

CLI reference

Every zavu command, flag by flag.

Define an agent

defineAgent API, providers, prompts, models.

Define tools

defineTool API, schemas, handlers, error handling.

Secrets

Environment variables encrypted at rest.

Triggers

Listen to Zavu events from your function.

Runtime versions

How pinning works, when to upgrade.

Examples

Complete restaurant booking agent, walked through.

Getting Started

Build agents

Send & receive

Operate

Developer tools

Integrations

Zavu Functions

Zavu Functions

When to use Functions

AI agent with real actions

Custom webhooks

Scheduled jobs

Event-driven automation

How it relates to AI Agents

Mental model

Quick tour

Continue

Quickstart

CLI reference

Define an agent

Define tools

Secrets

Triggers

Runtime versions

Examples

​Zavu Functions

​When to use Functions

AI agent with real actions

Custom webhooks

Scheduled jobs

Event-driven automation

​How it relates to AI Agents

​Mental model

​Quick tour

​Continue

Quickstart

CLI reference

Define an agent

Define tools

Secrets

Triggers

Runtime versions

Examples

Zavu Functions

When to use Functions

How it relates to AI Agents

Mental model

Quick tour

Continue