liveBook · Manning

Server agent

As explained in Context Engineering

A server agent is an A2A interaction role that exposes compatible endpoints, processes requests, and returns results.

A server agent is the receiving role in an agent-to-agent (A2A) interaction. It exposes compatible endpoints, processes a client agent’s request within its own isolated context, and returns results. The roles are dynamic: the same agent can act as a client when it delegates a subtask. section 7.8.1

An agent card lets clients discover the server’s capabilities before sending work. The card can describe its skills, supported interfaces, feature flags, and security requirements. A client uses that information to select an interface and submit task context. section 7.8.1 section 7.9.6

Simplified weather server

A Flask server publishes an agent card and accepts a message:send request. The client fetches the card, selects an interface, and sends a location in a message. The server extracts the location and returns a completed task-shaped response with an artifact; the weather values are randomly generated. section 7.9.6

Listing 7.7 Flask server snippet for handling a simplified A2A message request
@app.route('/message:send', methods=['POST'])
def send_message():
    request_payload = request.json or {}
    message = request_payload.get("message", {})
    parts = message.get("parts", [])

    location = "an unspecified location"
    for part in parts:
        text = part.get("text", "")
        if "location:" in text.lower():
            location = text.split(":", 1)[1].strip()
            break

    weather = random.choice(["Sunny", "Cloudy", "Windy"]) #1
    temperature = random.randint(30, 90)
    response_text = f"The current weather in {location} is {weather}
    at {temperature} degrees F."

    response_payload = {
        "task": {
            "id": str(uuid.uuid4()),
            "contextId": str(uuid.uuid4()),
            "status": {"state": "TASK_STATE_COMPLETED"},
            "artifacts": [
                {
                    "artifactId": str(uuid.uuid4()),
                    "name": "Current weather",
                    "parts": [{"text": response_text}]
                }
            ]
        }
    }
    return jsonify(response_payload), 200, {"Content-Type":
    "application/a2a+json"}

Inspect the server’s message handler, location extraction, and completed task-shaped response. section 7.9.6

The weather server is a teaching scaffold, not a conformant A2A implementation. Production systems should select the protocol binding from the agent card, validate capabilities, authenticate requests, handle task and context identifiers correctly, and support the required interaction mechanism. Consult the current specification because A2A continues to evolve. section 7.9.6 section 7.8.1

How Context Engineering covers it

How deep it goes
Explains discovery and message handling through a simplified server, while deferring full protocol conformance and production safeguards.
Examples
Uses a Flask weather server and client that discover an interface, exchange a location message, and return randomly generated weather.
What you can do after
Evaluate the server-agent interaction flow and identify safeguards missing from a simplified implementation.
How it connects
Connects agent cards, capability discovery, task context, protocol bindings, and artifacts to agent-to-agent communication.

Related concepts

This explainer was generated by Manning's knowledge graph from the text of Context Engineering. The section links point to the parts of the book it draws on.