Build a Gmail Agent in 5 Minutes with Invoke

mercury0100

— 2 min read
Build a Gmail Agent in 5 Minutes with Invoke

Managing OAuth, writing boilerplate, and wrestling with API quirks can turn a simple email agent into a days‑long project. With Invoke, you’re just a few lines of JSON and code away from a fully authenticated Gmail integration, no hidden hacks required.

Watch the full demo: YouTube Demo

Why Invoke?

  • Zero boilerplate OAuth: No manual client creation in Google Cloud Console—Invoke handles it for you.
  • Runtime discovery: Add or update integrations without restarting your agent.
  • Human‑friendly JSON: agents.json files are readable, versionable, and shareable.

Prerequisites

  1. Python 3.8+
  2. An OpenAI API key (or another compatible LLM)
  3. pip install invoke-agent

Quickstart: Built‑In Gmail Integration

Invoke ships with first‑class support for Google APIs. To get started, simply:

from langchain_openai import ChatOpenAI
from invoke_agent.agent import InvokeAgent

llm = ChatOpenAI(model="gpt-4.1")  
invoke = InvokeAgent(llm, agents=["google-calendar", "google-calendar"])

while True:
    user_input = input("📝 You: ").strip()
    if user_input.lower() in ["exit", "quit"]:
        break
    response = invoke.chat(user_input)
    print("\n🤖", response)

On first run, you’ll be prompted to authenticate via a link. Invoke stores tokens securely and refreshes them automatically, so you can focus on workflow.

Customizing with Your Own agents.json

Prefer full control? Download our Gmail agents.json template from the catalog. Edit the auth.code field:

  • "i" → Invoke‑managed OAuth (default)
  • null or "o" → Manual OAuth: you’ll supply client ID, secret, and scopes at runtime
{
  "agent": "google-gmail-send",
  "label": "📨 Gmail Send-Only API",
  "base_url": "https://www.googleapis.com",
  "auth": {
    "type": "oauth",
    "code": "o"
  },
  "endpoints": [
    {
      "name": "send_email",
      "label": "✉️ Send Email",
      "description": "Send an email via Gmail.",
      "method": "POST",
      "path": "/gmail/v1/users/me/messages/send",
      "body_params": {
        "raw": "The entire email content, base64‑encoded (RFC 2822 format)."
      },
      "notes": [
        "Construct 'to', 'subject', 'body' headers and body before encoding."
      ]
    }
  ]
}

Then point Invoke at your file:

from langchain_openai import ChatOpenAI
from invoke_agent.agent import InvokeAgent

llm = ChatOpenAI(model="gpt-4.1")  
invoke = InvokeAgent(llm, agents=["/path/to/agents.json"])

# Or specify a URL where the agents.json file is hosted
# invoke = InvokeAgent(llm, agents=["https://example.com/agents.json"])

while True:
    user_input = input("📝 You: ").strip()
    if user_input.lower() in ["exit", "quit"]:
        break
    response = invoke.chat(user_input)
    print("\n🤖", response)

How It Works Under the Hood

  1. Discovery: Invoke reads your agents list at startup and creates tool definitions for the LLM.
  2. Function Call: When the model issues a function call, Invoke maps it to the correct HTTP request.
  3. Auth Management: Tokens are obtained via OAuth flow or client‑secret input, then stored and refreshed.
  4. Execution & Parsing: Responses are returned to the agent, which parses JSON back into structured data.

This abstraction means your prompts stay concise, and you’re shielded from HTTP details and token lifecycles.

Next Steps

  • 🎯 Explore more: Check out prebuilt JSON for Slack, Notion, and Google Calendar in our catalog.
  • 📝 Contribute: Add your own agents.json to the marketplace.
  • 🚀 Deploy: Integrate your agent into production workflows, chatbots, or automation pipelines.

Ready to build? Clone the repo and start coding: https://github.com/mercury0100/invoke

Happy automating!

Read more