Gmail

This connector is optimized for AI agents. For the data replication connector, see Gmail.

The Gmail agent connector is a Python package that equips AI agents to interact with Gmail through strongly typed, well-documented tools. It's ready to use directly in your Python app, in an agent framework, or exposed through an MCP.

Gmail is Google's email service that provides email sending, receiving, and organization capabilities. This connector provides access to messages, threads, labels, drafts, and user profile information. It supports read operations for listing and retrieving email data, as well as write operations including sending messages, managing drafts, modifying message labels, and creating or updating labels.

Example prompts

The Gmail connector is optimized to handle prompts like these.

• List my recent emails
• Show me unread messages in my inbox
• Get the details of a specific email
• List all my Gmail labels
• Show me details for a specific label
• List my email drafts
• Get the content of a specific draft
• List my email threads
• Show me the full thread for a conversation
• Get my Gmail profile information
• Send an email to someone
• Create a new email draft
• Archive a message by removing the INBOX label
• Mark a message as read
• Mark a message as unread
• Move a message to trash
• Create a new label
• Update a label name or settings
• Delete a label
• Search for messages matching a query
• Find emails from a specific sender
• Show me emails with attachments

Unsupported prompts

The Gmail connector isn't currently able to handle prompts like these.

• Attach a file to an email
• Forward an email to someone
• Create a filter or rule
• Manage Gmail settings
• Access Google Calendar events
• Manage contacts

Entities and actions

This connector supports the following entities and actions. For more details, see this connector's full reference documentation.

Entity	Actions
Profile	Get, Context Store Search
Messages	List, Get, Create, Update, Context Store Search
Labels	List, Create, Get, Update, Delete, Context Store Search
Drafts	List, Create, Get, Update, Delete, Context Store Search
Drafts Send	Create
Threads	List, Get, Context Store Search
Messages Trash	Create
Messages Untrash	Create

Gmail API docs

See the official Gmail API reference.

SDK installation

uv pip install airbyte-agent-sdk

SDK usage

Connectors can run in hosted or open source mode.

Hosted

In hosted mode, API credentials are stored securely in Airbyte Agents. You provide your Airbyte credentials instead. If your Airbyte client can access multiple organizations, also set organization_id.

This example assumes you've already authenticated your connector with Airbyte. See Authentication to learn more about authenticating. If you need a step-by-step guide, see the hosted execution tutorial.

The connect() factory returns a fully typed GmailConnector and reads AIRBYTE_CLIENT_ID / AIRBYTE_CLIENT_SECRET from the environment:

Pydantic AI

Pydantic AI

from pydantic_ai import Agent
from airbyte_agent_sdk import connect
from airbyte_agent_sdk.connectors.gmail import GmailConnector

connector = connect("gmail", workspace_name="<your_workspace_name>")

agent = Agent("openai:gpt-4o")

@agent.tool_plain
@GmailConnector.tool_utils
async def gmail_execute(entity: str, action: str, params: dict | None = None):
    return await connector.execute(entity, action, params or {})

LangChain

LangChain

from langchain_core.tools import tool
from airbyte_agent_sdk import connect
from airbyte_agent_sdk.connectors.gmail import GmailConnector

connector = connect("gmail", workspace_name="<your_workspace_name>")

@tool
@GmailConnector.tool_utils
async def gmail_execute(entity: str, action: str, params: dict | None = None):
    """Execute Gmail connector operations."""
    result = await connector.execute(entity, action, params or {})
    # connector.execute returns a Pydantic envelope for typed actions; fall back to raw data otherwise.
    return result.model_dump(mode="json") if hasattr(result, "model_dump") else result

OpenAI Agents

OpenAI Agents

from agents import Agent, function_tool
from airbyte_agent_sdk import connect
from airbyte_agent_sdk.connectors.gmail import GmailConnector

connector = connect("gmail", workspace_name="<your_workspace_name>")

# strict_mode=False because `params: dict` is permissive and the default strict
# JSON schema rejects objects with additionalProperties.
@function_tool(strict_mode=False)
@GmailConnector.tool_utils(framework="openai_agents")
async def gmail_execute(entity: str, action: str, params: dict | None = None):
    """Execute Gmail connector operations."""
    result = await connector.execute(entity, action, params or {})
    return result.model_dump(mode="json") if hasattr(result, "model_dump") else result

agent = Agent(name="Gmail Assistant", tools=[gmail_execute])

FastMCP

FastMCP

from fastmcp import FastMCP
from airbyte_agent_sdk import connect
from airbyte_agent_sdk.connectors.gmail import GmailConnector

connector = connect("gmail", workspace_name="<your_workspace_name>")

mcp = FastMCP("Gmail Agent")

@mcp.tool
@GmailConnector.tool_utils
async def gmail_execute(entity: str, action: str, params: dict | None = None):
    """Execute Gmail connector operations."""
    result = await connector.execute(entity, action, params or {})
    return result.model_dump(mode="json") if hasattr(result, "model_dump") else result

Or pass credentials explicitly (equivalent, useful when you're not loading them from the environment):

Pydantic AI

Pydantic AI

from pydantic_ai import Agent
from airbyte_agent_sdk.connectors.gmail import GmailConnector
from airbyte_agent_sdk.types import AirbyteAuthConfig

connector = GmailConnector(
    auth_config=AirbyteAuthConfig(
        workspace_name="<your_workspace_name>",
        organization_id="<your_organization_id>",  # Optional for multi-org clients
        airbyte_client_id="<your-client-id>",
        airbyte_client_secret="<your-client-secret>"
    )
)

agent = Agent("openai:gpt-4o")

@agent.tool_plain
@GmailConnector.tool_utils
async def gmail_execute(entity: str, action: str, params: dict | None = None):
    return await connector.execute(entity, action, params or {})

LangChain

LangChain

from langchain_core.tools import tool
from airbyte_agent_sdk.connectors.gmail import GmailConnector
from airbyte_agent_sdk.types import AirbyteAuthConfig

connector = GmailConnector(
    auth_config=AirbyteAuthConfig(
        workspace_name="<your_workspace_name>",
        organization_id="<your_organization_id>",  # Optional for multi-org clients
        airbyte_client_id="<your-client-id>",
        airbyte_client_secret="<your-client-secret>"
    )
)

@tool
@GmailConnector.tool_utils
async def gmail_execute(entity: str, action: str, params: dict | None = None):
    """Execute Gmail connector operations."""
    result = await connector.execute(entity, action, params or {})
    # connector.execute returns a Pydantic envelope for typed actions; fall back to raw data otherwise.
    return result.model_dump(mode="json") if hasattr(result, "model_dump") else result

OpenAI Agents

OpenAI Agents

from agents import Agent, function_tool
from airbyte_agent_sdk.connectors.gmail import GmailConnector
from airbyte_agent_sdk.types import AirbyteAuthConfig

connector = GmailConnector(
    auth_config=AirbyteAuthConfig(
        workspace_name="<your_workspace_name>",
        organization_id="<your_organization_id>",  # Optional for multi-org clients
        airbyte_client_id="<your-client-id>",
        airbyte_client_secret="<your-client-secret>"
    )
)

# strict_mode=False because `params: dict` is permissive and the default strict
# JSON schema rejects objects with additionalProperties.
@function_tool(strict_mode=False)
@GmailConnector.tool_utils(framework="openai_agents")
async def gmail_execute(entity: str, action: str, params: dict | None = None):
    """Execute Gmail connector operations."""
    result = await connector.execute(entity, action, params or {})
    return result.model_dump(mode="json") if hasattr(result, "model_dump") else result

agent = Agent(name="Gmail Assistant", tools=[gmail_execute])

FastMCP

FastMCP

from fastmcp import FastMCP
from airbyte_agent_sdk.connectors.gmail import GmailConnector
from airbyte_agent_sdk.types import AirbyteAuthConfig

connector = GmailConnector(
    auth_config=AirbyteAuthConfig(
        workspace_name="<your_workspace_name>",
        organization_id="<your_organization_id>",  # Optional for multi-org clients
        airbyte_client_id="<your-client-id>",
        airbyte_client_secret="<your-client-secret>"
    )
)

mcp = FastMCP("Gmail Agent")

@mcp.tool
@GmailConnector.tool_utils
async def gmail_execute(entity: str, action: str, params: dict | None = None):
    """Execute Gmail connector operations."""
    result = await connector.execute(entity, action, params or {})
    return result.model_dump(mode="json") if hasattr(result, "model_dump") else result

Open source

In open source mode, you provide API credentials directly to the connector.

Pydantic AI

Pydantic AI

from pydantic_ai import Agent
from airbyte_agent_sdk.connectors.gmail import GmailConnector
from airbyte_agent_sdk.connectors.gmail.models import GmailAuthConfig

connector = GmailConnector(
    auth_config=GmailAuthConfig(
        access_token="<Your Google OAuth2 Access Token (optional, will be obtained via refresh)>",
        refresh_token="<Your Google OAuth2 Refresh Token>",
        client_id="<Your Google OAuth2 Client ID>",
        client_secret="<Your Google OAuth2 Client Secret>"
    )
)

agent = Agent("openai:gpt-4o")

@agent.tool_plain
@GmailConnector.tool_utils
async def gmail_execute(entity: str, action: str, params: dict | None = None):
    return await connector.execute(entity, action, params or {})

LangChain

LangChain

from langchain_core.tools import tool
from airbyte_agent_sdk.connectors.gmail import GmailConnector
from airbyte_agent_sdk.connectors.gmail.models import GmailAuthConfig

connector = GmailConnector(
    auth_config=GmailAuthConfig(
        access_token="<Your Google OAuth2 Access Token (optional, will be obtained via refresh)>",
        refresh_token="<Your Google OAuth2 Refresh Token>",
        client_id="<Your Google OAuth2 Client ID>",
        client_secret="<Your Google OAuth2 Client Secret>"
    )
)

@tool
@GmailConnector.tool_utils
async def gmail_execute(entity: str, action: str, params: dict | None = None):
    """Execute Gmail connector operations."""
    result = await connector.execute(entity, action, params or {})
    # connector.execute returns a Pydantic envelope for typed actions; fall back to raw data otherwise.
    return result.model_dump(mode="json") if hasattr(result, "model_dump") else result

OpenAI Agents

OpenAI Agents

from agents import Agent, function_tool
from airbyte_agent_sdk.connectors.gmail import GmailConnector
from airbyte_agent_sdk.connectors.gmail.models import GmailAuthConfig

connector = GmailConnector(
    auth_config=GmailAuthConfig(
        access_token="<Your Google OAuth2 Access Token (optional, will be obtained via refresh)>",
        refresh_token="<Your Google OAuth2 Refresh Token>",
        client_id="<Your Google OAuth2 Client ID>",
        client_secret="<Your Google OAuth2 Client Secret>"
    )
)

# strict_mode=False because `params: dict` is permissive and the default strict
# JSON schema rejects objects with additionalProperties.
@function_tool(strict_mode=False)
@GmailConnector.tool_utils(framework="openai_agents")
async def gmail_execute(entity: str, action: str, params: dict | None = None):
    """Execute Gmail connector operations."""
    result = await connector.execute(entity, action, params or {})
    return result.model_dump(mode="json") if hasattr(result, "model_dump") else result

agent = Agent(name="Gmail Assistant", tools=[gmail_execute])

FastMCP

FastMCP

from fastmcp import FastMCP
from airbyte_agent_sdk.connectors.gmail import GmailConnector
from airbyte_agent_sdk.connectors.gmail.models import GmailAuthConfig

connector = GmailConnector(
    auth_config=GmailAuthConfig(
        access_token="<Your Google OAuth2 Access Token (optional, will be obtained via refresh)>",
        refresh_token="<Your Google OAuth2 Refresh Token>",
        client_id="<Your Google OAuth2 Client ID>",
        client_secret="<Your Google OAuth2 Client Secret>"
    )
)

mcp = FastMCP("Gmail Agent")

@mcp.tool
@GmailConnector.tool_utils
async def gmail_execute(entity: str, action: str, params: dict | None = None):
    """Execute Gmail connector operations."""
    result = await connector.execute(entity, action, params or {})
    return result.model_dump(mode="json") if hasattr(result, "model_dump") else result

Authentication

For all authentication options, see the connector's authentication documentation.

Version information

Connector version: 0.1.4