Quick Start

What is A2E?

A2E (Agent-to-Environment) is a protocol and Python runtime for building stateful, interactive environments that LLM agents can interact with. It standardizes how agents use tools, access memory, run processes, observe environments, and learn from feedback — enabling a shift from static prompting to dynamic interaction.

Think of it as "POSIX for AI agents" — a unified interface between intelligent agents and the environments they operate in.

Install

bash

pip install a2e

Minimal Server

Create a config.yaml:

yaml

host_id: "my-a2e-host"
server:
  host: "0.0.0.0"
  port: 8765
auth_token: "dev-secret"
transport:
  type: http
audit:
  enabled: true
  path: "/tmp/a2e-audit.jsonl"
plugins:
  - name: mytools
    type: tools
    cls: cookbook.servers.tools.registry_tool_plugin.RegistryToolPlugin
    metadata:
      enabled: true
      modules:
        - cookbook.servers.tools.read_file_tool
        - cookbook.servers.tools.glob_tool
  - name: mymemory
    type: memory
    cls: cookbook.servers.memory.inmemory.InMemoryPlugin
    metadata:
      enabled: true
      working_limit: 50
      episodic_limit: 50
      semantic_limit: 50

Start the server:

python

from a2e.schema import A2EHostConfig
from a2e.core.server.server import A2EServer

config = A2EHostConfig.from_yaml("config.yaml")
server = A2EServer(config)
app = server.start()  # Returns FastAPI app for HTTP mode

import uvicorn
uvicorn.run(app, host="0.0.0.0", port=8765)

Minimal Client

python

from a2e.core.client.client import A2EClient
from a2e.core.transports import HTTPTransport, HTTPTransportConfig, build_transport
from a2e.caps.tools.client import ToolAPI
from a2e.caps.memory.client import MemoryAPI

# Build transport and connect
transport = build_transport(config.transport, logger=None)
client = A2EClient(
    transport=transport,
    agent_id="my-agent",
    auth_token="dev-secret",
    agent_caps=["tools", "memory"]
)

with client:
    # Use capability APIs
    tools = ToolAPI(client)
    memory = MemoryAPI(client)

    # List and call tools
    tool_list = tools.list()
    result = tools.call("read_file", {"path": "/etc/hostname"})

    # Store and retrieve memory
    memory.remember("user_name", "Alice", tier="working")
    name = memory.recall("user_name")  # Returns "Alice"

What Happens Behind the Scenes

Transport starts — HTTP client connects to the server
Handshake — Client sends handshake/req with agent_caps, server responds with accepted_caps
RPC calls — Each tools.call() sends a tool/call/req, gets back tool/call/resp (with streaming events via tool/event)
Disconnect — Context manager sends shutdown and stops the transport

Quick Start ​

What is A2E? ​

Install ​

Minimal Server ​

Minimal Client ​

What Happens Behind the Scenes ​

Next Steps ​