appif 0.3.0

Messaging connectors giving AI agents authenticated, normalized access to email and chat platforms

appif -- Application Interfaces

A Python library of messaging connectors that give AI agents authenticated, normalized access to email and chat platforms.

Purpose

Agents need information that lives behind logins: email threads, Slack messages, team conversations. This library provides a set of connectors that authenticate as you and return clean, structured MessageEvent objects suitable for agent reasoning -- platform-specific APIs are fully encapsulated behind a shared Connector protocol.

For the complete usage guide -- the unified model, per-connector mapping tables, code examples, and environment variable reference -- see docs/usage.md.

Quick Start

pip install appif

from appif.adapters.gmail import GmailConnector
from appif.domain.messaging.models import MessageEvent, MessageContent

class MyListener:
    def on_message(self, event: MessageEvent) -> None:
        # Every platform delivers the same MessageEvent:
        #   event.author.display_name  -- who sent it
        #   event.content.text         -- what they said
        #   event.conversation_ref     -- opaque handle for reply
        #   event.metadata             -- platform-specific extras
        print(f"[{event.connector}] {event.author.display_name}: {event.content.text}")

connector = GmailConnector()
connector.connect()
connector.register_listener(MyListener())

# Reply to any received message:
# connector.send(event.conversation_ref, MessageContent(text="Got it."))

All three connectors (Gmail, Outlook, Slack) follow this same pattern. The full model, per-connector setup, and examples are in docs/usage.md.

Supported Connectors

Service	Connector	Inbound Method	Status
Gmail	Google API (OAuth 2.0)	history.list polling	Active
Outlook / Microsoft 365	Microsoft Graph API	Delta-query polling	Active
Slack	Slack API (Bolt + Socket Mode)	Real-time Socket Mode	Active

See docs/usage.md for how each platform maps to the unified model, setup instructions, and working code examples.

Installation

For development

uv venv .venv
source .venv/bin/activate
uv pip install -e ".[dev]"

As a library dependency

pip install appif

Prerequisites

• Python 3.13.x
• uv (for development)

Configuration

Credentials are stored in ~/.env and loaded at runtime:

cp .env.example ~/.env

Variable	Service	Required
APPIF_GMAIL_CLIENT_ID	Gmail	Yes -- Google Cloud OAuth client ID
APPIF_GMAIL_CLIENT_SECRET	Gmail	Yes -- Google Cloud OAuth client secret
APPIF_GMAIL_ACCOUNT	Gmail	Yes -- Account email address
APPIF_OUTLOOK_CLIENT_ID	Outlook	Yes -- Azure AD app (client) ID
APPIF_OUTLOOK_TENANT_ID	Outlook	Optional -- Azure AD tenant (default: common)
APPIF_SLACK_BOT_OAUTH_TOKEN	Slack	Yes -- Bot user OAuth token (xoxb-...)
APPIF_SLACK_BOT_APP_LEVEL_TOKEN	Slack	Yes -- App-level token for Socket Mode (xapp-...)

See .env.example for the full template with all optional variables.

Project Structure

application_interfaces/
├── src/
│   └── appif/                       # Top-level package (PyPI: appif)
│       ├── __init__.py              # Version via importlib.metadata
│       ├── domain/
│       │   └── messaging/           # Connector protocol, canonical models, errors
│       ├── adapters/
│       │   ├── gmail/               # Gmail messaging connector
│       │   ├── outlook/             # Outlook messaging connector
│       │   └── slack/               # Slack messaging connector
│       └── infrastructure/          # Credential loading
├── tests/
│   ├── unit/                        # 241 unit tests
│   ├── integration/
│   └── e2e/
├── scripts/                         # OAuth consent flows
├── docs/design/                     # Design documents per adapter
├── pyproject.toml
├── ADAPTERS.md                      # Detailed adapter documentation
├── .env.example
└── readme.md

Development

# Set up dev environment
uv venv .venv
source .venv/bin/activate
uv pip install -e ".[dev]"

# Run all unit tests (241 tests)
pytest tests/unit -v

# Run adapter-specific tests
pytest tests/unit/test_gmail_*.py -v
pytest tests/unit/test_outlook_*.py -v

# Lint and format
ruff check src/ tests/
ruff format src/ tests/

# Type check
mypy src/

Architecture

Connector Protocol

All messaging connectors implement a shared Connector protocol (appif.domain.messaging.ports.Connector) -- a transport adapter that:

• Connects to an external system and manages authentication
• Emits normalized MessageEvent objects to registered listeners
• Delivers outbound messages via send(target, content)
• Supports historical backfill alongside realtime event ingestion
• Advertises capabilities so upstream logic branches on what the connector supports, not which platform it is

All connectors produce identical canonical types (MessageEvent, ConversationRef, SendReceipt). Platform-specific SDK code is fully encapsulated -- zero Slack/Outlook/Gmail types leak through the public interface.

Internal Module Pattern

Each adapter follows the same decomposition:

src/appif/adapters/<platform>/
├── __init__.py          # Public exports
├── connector.py         # Connector protocol implementation
├── _auth.py             # Authentication (protocol + implementation)
├── _normalizer.py       # Platform message -> MessageEvent
├── _message_builder.py  # MessageContent -> platform request (email adapters)
├── _poller.py           # Inbound message detection (email adapters)
└── _rate_limiter.py     # Retry + platform error -> domain error mapping

Credential Setup

Adapter	Consent Script	Setup Guide
Gmail	python scripts/gmail_consent.py <account>	docs/design/gmail/setup.md
Outlook	python scripts/outlook_consent.py <account>	docs/design/outlook/setup.md
Slack	N/A (tokens from Slack app)	docs/design/slack/setup.md

License

MIT -- see LICENSE.