Agents & protocols

Agents are managed at runtime from the built-in admin panel (/admin) and stored in PostgreSQL — no redeploy is needed to add one. For each agent you choose its protocol, its endpoint, and who can reach it.

Supported protocols

Agentgram speaks three protocols on the way in and converts all of them to AG-UI events on the way out:

Protocol	Transport	Contract
Custom	REST endpoint returning SSE or JSON	REST contract
A2A	JSON-RPC (Agent-to-Agent)	A2A contract
ADK	Google ADK over REST/SSE	ADK contract

See the protocols overview for how each one is mapped onto AG-UI.

Permissions (RBAC)

Access is role-based and controlled per agent (and, the same way, per MCP server). An agent can be reachable by:

Groups — e.g. Google Workspace groups (google-workspace/team@example.com), with * for public access.
Individual users — by email.

When a request comes in, the API filters the visible agents by the caller’s identity (from the JWT claims) before routing — so a single endpoint serves your whole company, yet each person only ever reaches the agents and MCP servers they’re entitled to.

The chat contract

A chat request is a list of messages plus an optional session id:

POST /api/agents/{agentId}/chat
Content-Type: application/json
Authorization: Bearer <jwt>      # only when auth is enabled

{
  "messages": [
    { "role": "user", "content": "Hello" }
  ],
  "session_id": "optional-uuid"
}

The response is a stream of AG-UI events over SSE — the same shape regardless of the agent’s protocol:

data: {"type":"RUN_STARTED","threadId":"...","runId":"..."}
data: {"type":"TEXT_MESSAGE_START","messageId":"...","role":"assistant"}
data: {"type":"TEXT_MESSAGE_CONTENT","messageId":"...","delta":"Hello"}
data: {"type":"TEXT_MESSAGE_END","messageId":"..."}
data: {"type":"RUN_FINISHED","threadId":"...","runId":"..."}

Full endpoint details are in the API Reference.

Configuration Sharing & collaboration