Documentation Index
Fetch the complete documentation index at: https://agno-v2-docs-align-with-readme.mintlify.app/llms.txt
Use this file to discover all available pages before exploring further.
An interface is an adapter between AgentOS and a surface where users live. It receives events from Slack, Telegram, WhatsApp, or other channels, maps them onto AgentOS’s run and session model, and routes responses back to the right thread, channel, or user.
The agent doesn’t change. The same Agent instance answers a Slack DM, a Telegram message, and an AG-UI browser stream. Memory carries across all of them, scoped per (user_id, session_id).
Available interfaces
Two categories. Chat surfaces meet humans where they already are. Protocol surfaces are how other systems talk to your agent.
Chat surfaces
| Interface | Use case | Setup |
|---|
| Slack | Team chat, DMs, channel mentions, thread sessions | Slack |
| Telegram | Personal assistants, mobile chat | Telegram |
| WhatsApp | Customer support, mobile chat with E2E encryption support | WhatsApp |
| Discord | Community servers, gaming, custom commands. Runs in its own process via agno.integrations.discord. | Discord |
Protocol surfaces
| Interface | Use case | Setup |
|---|
| A2A | Other agents talk to yours over a standardized agent-to-agent protocol | A2A |
| AG-UI | Browser clients consuming SSE streams of run output | AG-UI |
Setup
Each interface registers its own routes on the FastAPI app. Slack lands events at /slack/events. Telegram at /telegram/webhook. The agent=... parameter tells the interface which agent to dispatch incoming messages to.
from agno.os import AgentOS
from agno.os.interfaces.slack import Slack
from agno.os.interfaces.telegram import Telegram
agent_os = AgentOS(
agents=[agent],
db=db,
interfaces=[
Slack(agent=agent, token="xoxb-...", signing_secret="..."),
Telegram(agent=agent, token="bot-token"),
],
)
If your AgentOS has multiple agents, you can wire each interface to a different one — Slack to your customer support agent, Telegram to a personal assistant — or wire several interfaces to the same agent.
Credentials at a glance
Per-interface setup pages have the full OAuth flows, scope lists, and webhook configuration. The summary:
| Interface | Needs |
|---|
| Slack | Bot token (xoxb-...), signing secret, OAuth scopes for the events you handle |
| Telegram | Bot token from @BotFather |
| WhatsApp | Business API token, verify token, phone number ID |
| Discord | Bot token, application ID |
| A2A | Inbound JWTs validated by AgentOS auth |
| AG-UI | AgentOS JWT in the SSE request |
Sessions per surface
Every interface maps surface state to AgentOS sessions, so a conversation in Slack carries forward like any other session — the next reply in the same thread reuses the same session and history, no re-mentioning required.
| Interface | Session ID | User ID |
|---|
| Slack | Thread timestamp | Slack user ID |
| Telegram | Chat ID | Telegram user ID |
| WhatsApp | Phone number | Phone number |
| Discord | Channel + thread | Discord user ID |
| A2A | Caller-supplied | Caller-supplied |
| AG-UI | Browser session | JWT subject |
Resolving Slack user IDs to names
By default Slack hands you opaque user IDs like U07ABCXYZ. If your agent’s responses or memory are clearer when grounded in human names, set resolve_user_identity=True on the Slack interface:
Slack(agent=agent, token=..., signing_secret=..., resolve_user_identity=True)
The interface looks up each user ID against users.info and passes the resolved name through to the agent. Off by default because it costs an extra Slack API call per message and isn’t always wanted.
One agent, many surfaces
A single agent can answer on every surface at once. Memory follows the user across surfaces — assuming you can map their Slack ID to the same user_id the AG-UI client passes:
agent_os = AgentOS(
agents=[support_agent],
db=db,
interfaces=[
Slack(agent=support_agent, token=..., signing_secret=...),
Telegram(agent=support_agent, token=...),
WhatsApp(agent=support_agent, token=..., verify_token=...),
AGUI(agent=support_agent),
],
)
The same questions the agent answered in Slack last week show up in its memory when the user opens the AG-UI widget on your website. The agent doesn’t know which surface a request came from. It just sees a user_id, a session_id, and a message.
Conditional registration
Don’t register interfaces you don’t have credentials for:
interfaces = []
if SLACK_TOKEN and SLACK_SIGNING_SECRET:
interfaces.append(Slack(agent=agent, token=SLACK_TOKEN, signing_secret=SLACK_SIGNING_SECRET))
if TELEGRAM_TOKEN:
interfaces.append(Telegram(agent=agent, token=TELEGRAM_TOKEN))
agent_os = AgentOS(agents=[agent], db=db, interfaces=interfaces)
The Scout, Dash, and Coda tutorials use this pattern. Slack only loads when both env vars are set, so dev runs without credentials don’t crash.
Custom interfaces and one-off webhooks
The interface API is small. Subclass BaseInterface, register routes on the FastAPI app via setup, dispatch incoming messages to the agent. See BaseInterface for the full surface.
For one-off webhooks (a CRM event, a GitHub action, a custom dashboard), don’t write an interface. Add a route directly to the FastAPI app:
app = agent_os.get_app()
@app.post("/webhooks/stripe")
async def handle_stripe(event: dict):
response = await agent.arun(f"Process Stripe event: {event}", user_id="system")
return {"ok": True, "response": response.content}
A custom interface is for surfaces you’ll reuse across agents and OS instances. A direct route is for one-off integrations.
Next
Scheduling →