Agent MessengerWhat is Agent Messenger?
Give your AI agent the power to read and send messages across Slack, Discord, Teams, Telegram, Channel Talk and more
Agent Messenger is a unified, agent-friendly CLI for messaging platforms. Zero-config credential extraction from your desktop apps—no OAuth flows, no API keys, no admin approval needed. Works out of the box. Supports Slack, Discord, Teams, Telegram, and Channel Talk (beta).
Why Agent Messenger?
Messaging platforms only offer Bot tokens for API access—your AI agent can never act as you. Agent Messenger extracts user credentials directly from your installed desktop apps, letting your agent operate on your behalf. Bot tokens are fully supported too for server-side and CI/CD use cases.
- Act as yourself, not a bot — Extracted user tokens let your agent operate on your behalf
- No API keys needed — Automatically extracts credentials from your installed desktop apps
- One interface, multiple platforms — Learn once, use everywhere (Slack, Discord, Teams, Telegram, Channel Talk)
- AI-agent friendly — JSON output by default, perfect for LLM tool use
- Human friendly too — Add
--prettyfor readable output - Token efficient — CLI, not MCP. Load only what you need
How It Works
Agent Messenger reads authentication tokens directly from your installed desktop apps (Slack, Discord, Teams, Channel Talk). Telegram uses TDLib with auto-provisioned API credentials. These apps store session tokens locally, and Agent Messenger safely extracts them.
# Extract credentials from Slack desktop app
agent-slack auth extract
# That's it! Now you can send messages
agent-slack message send general "Hello from my AI agent!"No app creation. No admin approval. No waiting.
Key Features
| Feature | Description |
|---|---|
| Auto-extraction | Pulls tokens from desktop apps automatically |
| Multi-platform | Slack, Discord, Microsoft Teams, Telegram, and Channel Talk (beta) |
| Multi-workspace | Switch between workspaces/servers easily |
| JSON output | Machine-readable by default, --pretty for humans |
| Comprehensive | Messages, channels, users, reactions, files |
| Bot support | agent-slackbot, agent-discordbot, and agent-channeltalkbot for server-side and CI/CD |
Why Not MCP?
The Model Context Protocol (MCP) is a popular way to give AI agents access to external tools. But it has a fundamental problem: context bloat.
MCP servers expose all their tools at once. When you connect an MCP server with 50 tools, your AI agent's context window fills up with 50 tool descriptions—even if it only needs one.
Agent Skills + CLI is better:
| MCP Approach | Agent Skills Approach |
|---|---|
| All tools loaded at once | Load only what you need |
| Bloated context window | Minimal token usage |
| Agent confused by options | Focused, relevant tools |
| Static tool definitions | Dynamic skill loading |
With Agent Messenger, your AI agent loads the skill it needs, uses the CLI, and moves on. No wasted tokens.
Why Not OAuth?
Traditional Slack/Discord/Teams integrations require OAuth:
- Create an app in the platform's developer portal
- Configure OAuth scopes and redirect URLs
- Submit for workspace admin approval
- Wait days (or weeks) for approval
- Handle token refresh, storage, and rotation
Agent Messenger skips all of this.
Your desktop apps already have valid session tokens. Agent Messenger extracts them directly, so you can start sending messages immediately.
When OAuth Makes Sense
OAuth is better when you need:
- Server-side bots - Running 24/7 without your desktop app
- Multi-user applications - Many users authenticating separately
- Granular permissions - Fine-tuned access control
For these use cases, Agent Messenger also supports bot tokens via agent-slackbot, agent-discordbot, and agent-channeltalkbot.
Design Principles
Zero Configuration
The best setup is no setup. Agent Messenger should work the moment you install it.
AI-First, Human-Friendly
Default output is JSON for machines. Add --pretty for humans. Both are first-class citizens.
Unified Interface
Learn once, use everywhere. The same commands work across Slack, Discord, Teams, Telegram, and Channel Talk.
Fail Fast, Fail Clear
When something goes wrong, tell the user exactly what and how to fix it. No cryptic error codes.
Respect Privacy
Tokens stay on your machine. Nothing is sent to external servers. Your credentials are yours.
Next Steps
- Quick Start - Get running in 30 seconds
- Slack - Full Slack command reference
- Slack Bot - Bot token integration for server-side and CI/CD
- Discord - Full Discord command reference
- Discord Bot - Bot token integration for server-side and CI/CD
- Teams - Full Teams command reference
- Telegram - TDLib setup and Telegram command reference
- Channel Talk - Full Channel Talk command reference (beta)
- Channel Talk Bot - Bot API integration for Channel Talk (beta)
Inspiration
Agent Messenger is inspired by agent-browser from Vercel Labs, which takes a similar approach to browser automation for AI agents.