Skip to main content
When building your agent, you can configure certain attributes that affect how it appears and behaves in the user interface. The @server.agent decorator accepts a detail parameter with an AgentDetail object that controls the visual representation and behavior in the UI. You can customize various aspects of your agent’s presentation, such as:
  • The type of user interface the agent uses (chat, form, etc.)
  • Custom user greetings
  • Metadata about tools the agent provides
  • Author and contributor information
  • License, framework, and source code details

​
Basic Configuration

Configuring agent details is straightforward. Import AgentDetail and related types, then pass them to the detail parameter in the @server.agent decorator: 
import os

from a2a.types import Message
from agentstack_sdk.server import Server
from agentstack_sdk.server.context import RunContext
from agentstack_sdk.a2a.extensions import AgentDetail, AgentDetailContributor, AgentDetailTool

server = Server()

@server.agent(
    name="Example Research Assistant",
    detail=AgentDetail(
        interaction_mode="multi-turn", # or single-turn
        user_greeting="Hi there! I can help you research topics or summarize uploaded documents.",
        tools=[
            AgentDetailTool(
                name="Web Search",
                description="Looks up recent and relevant information from the web."
            ),
            AgentDetailTool(
                name="Document Reader",
                description="Reads and extracts key insights from uploaded PDFs or text files."
            ),
        ],
        framework="BeeAI Framework",
        author={
            "name": "Agent Stack Team",
            "email": "team@example.com",
        },
        source_code_url="https://github.com/example/example-research-assistant",
    ),
    skills=[
        AgentSkill(
            id="research",
            name="Research",
            description=dedent(
                """\
                Finds up-to-date information on a given topic, synthesizes key points, 
                and summarizes findings in clear, useful responses.
                """
            ),
            tags=["Search", "Knowledge"],
            examples=[
                "Find recent news about AI ethics in 2025.",
                "What are the main challenges in renewable energy adoption?",
                "Give me an overview of current space exploration missions.",
            ],
        ),
        AgentSkill(
            id="summarization",
            name="Summarization",
            description=dedent(
                """\
                Reads and summarizes uploaded text or documents, highlighting the 
                most important ideas, statistics, and conclusions.
                """
            ),
            tags=["Documents", "Summaries"],
            examples=[
                "Summarize this PDF report about electric vehicle trends.",
                "What are the main points from this research article?",
                "Condense this document into a short summary I can share.",
            ],
        ),
    ],
)
async def example_agent(input: Message, context: RunContext):
    """An example agent with detailed configuration"""
    yield "Hello World!"

def run():
    server.run(host=os.getenv("HOST", "127.0.0.1"), port=int(os.getenv("PORT", 8000)))


if __name__ == "__main__":
    run()

​
Key Fields

​
UI Configuration

  • interaction_mode: Set to "multi-turn" for conversational agents or "single-turn" for one-shot agents
  • user_greeting: Welcome message displayed when users first interact with your agent

​
Metadata

  • framework: The framework used to build your agent (e.g., “BeeAI”, “LangGraph”, “CrewAI”)
  • programming_language: Primary language (e.g., “Python”, “JavaScript”)
  • license: Software license (e.g., “Apache 2.0”, “MIT”)

​
Tools

  • tools: List of AgentDetailTool objects describing your agent’s capabilities
    • Each tool has a name and description to help users understand each tool’s purpose

​
Skills

  • skills: List of AgentSkill objects that represent distinct, user-facing abilities your agent can perform
    • Each skill includes an id, name, description, tags, and examples
    • Examples are displayed in the user interface as starter questions — they help users understand what kinds of requests your agent can handle and encourage exploration
    • Skills define what your agent can do — for example, “Research” for finding and summarizing information, or “Summarization” for condensing long documents
  • source_code_url: Link to the source code repository
  • author: Primary author information (name, email, url)
  • contributors: List of additional contributors
The Agent Details extension helps users understand your agent’s capabilities before they start using it. Provide clear descriptions and accurate tool information to improve the user experience.