Agent as Judge with Teams

from typing import Optional

from agno.agent import Agent
from agno.db.sqlite import SqliteDb
from agno.eval.agent_as_judge import AgentAsJudgeEval, AgentAsJudgeResult
from agno.models.openai import OpenAIResponses
from agno.team.team import Team

# Setup database to persist eval results
db = SqliteDb(db_file="tmp/agent_as_judge_team.db")

# Setup a team with researcher and writer
researcher = Agent(
    name="Researcher",
    role="Research and gather information",
    model=OpenAIResponses(id="gpt-5.2"),
)

writer = Agent(
    name="Writer",
    role="Write clear and concise summaries",
    model=OpenAIResponses(id="gpt-5.2"),
)

research_team = Team(
    name="Research Team",
    model=OpenAIResponses(id="gpt-5.2"),
    members=[researcher, writer],
    instructions=["First research the topic thoroughly, then write a clear summary."],
    db=db,
)

response = research_team.run("Explain quantum computing")

evaluation = AgentAsJudgeEval(
    name="Team Response Quality",
    model=OpenAIResponses(id="gpt-5.2"),
    criteria="Response should be well-researched, clear, and comprehensive with good flow",
    scoring_strategy="binary",
    db=db,
)

result: Optional[AgentAsJudgeResult] = evaluation.run(
    input="Explain quantum computing",
    output=str(response.content),
    print_results=True,
    print_summary=True,
)

# Query database for stored results
print("Database Results:")
eval_runs = db.get_eval_runs()
print(f"Total evaluations stored: {len(eval_runs)}")
if eval_runs:
    latest = eval_runs[-1]
    print(f"Eval ID: {latest.run_id}")
    print(f"Team: {research_team.name}")