Tool Call Compression With Manager

Demonstrates custom tool result compression using CompressionManager.

tool_call_compression_with_manager.py
"""
Tool Call Compression With Manager
==================================

Demonstrates custom tool result compression using CompressionManager.
"""

from textwrap import dedent

from agno.agent import Agent
from agno.compression.manager import CompressionManager
from agno.db.sqlite import SqliteDb
from agno.models.openai import OpenAIResponses
from agno.team import Team
from agno.tools.websearch import WebSearchTools

# ---------------------------------------------------------------------------
# Setup
# ---------------------------------------------------------------------------
compression_prompt = """
    You are a compression expert. Your goal is to compress web search results for a competitive intelligence analyst.

    YOUR GOAL: Extract only actionable competitive insights while being extremely concise.

    MUST PRESERVE:
    - Competitor names and specific actions (product launches, partnerships, acquisitions, pricing changes)
    - Exact numbers (revenue, market share, growth rates, pricing, headcount)
    - Precise dates (announcement dates, launch dates, deal dates)
    - Direct quotes from executives or official statements
    - Funding rounds and valuations

    MUST REMOVE:
    - Company history and background information
    - General industry trends (unless competitor-specific)
    - Analyst opinions and speculation (keep only facts)
    - Detailed product descriptions (keep only key differentiators and pricing)
    - Marketing fluff and promotional language

    OUTPUT FORMAT:
    Return a bullet-point list where each line follows this format:
    "[Company Name] - [Date]: [Action/Event] ([Key Numbers/Details])"

    Keep it under 200 words total. Be ruthlessly concise. Facts only.

    Example:
    - Acme Corp - Mar 15, 2024: Launched AcmeGPT at $99/user/month, targeting enterprise market
    - TechCo - Feb 10, 2024: Acquired DataStart for $150M, gaining 500 enterprise customers
"""

compression_manager = CompressionManager(
    model=OpenAIResponses(id="gpt-5.2"),
    compress_tool_results_limit=2,  # Keep only last 2 tool call results uncompressed
    compress_tool_call_instructions=compression_prompt,
)

# ---------------------------------------------------------------------------
# Create Members
# ---------------------------------------------------------------------------
tech_researcher = Agent(
    name="Alex",
    role="Technology Researcher",
    model=OpenAIResponses(id="gpt-5.2"),
    instructions=dedent("""
        You specialize in technology and AI research.
        - Focus on latest developments, trends, and breakthroughs
        - Provide concise, data-driven insights
        - Cite your sources
    """).strip(),
)

business_analyst = Agent(
    name="Sarah",
    role="Business Analyst",
    model=OpenAIResponses(id="gpt-5.2"),
    instructions=dedent("""
        You specialize in business and market analysis.
        - Focus on companies, markets, and economic trends
        - Provide actionable business insights
        - Include relevant data and statistics
    """).strip(),
)

# ---------------------------------------------------------------------------
# Create Team
# ---------------------------------------------------------------------------
research_team = Team(
    name="Research Team",
    model=OpenAIResponses(id="gpt-5.2"),
    members=[tech_researcher, business_analyst],
    tools=[WebSearchTools()],  # Team uses DuckDuckGo for research
    description="Research team that investigates topics and provides analysis.",
    instructions=dedent("""
        You are a research coordinator that investigates topics comprehensively.

        Your Process:
        1. Use DuckDuckGo to search for a lot of information on the topic.
        2. Delegate detailed analysis to the appropriate specialist
        3. Synthesize research findings with specialist insights

        Guidelines:
        - Always start with web research using your DuckDuckGo tools. Try to get as much information as possible.
        - Choose the right specialist based on the topic (tech vs business)
        - Combine your research with specialist analysis
        - Provide comprehensive, well-sourced responses
    """).strip(),
    db=SqliteDb(db_file="tmp/research_team.db"),
    show_members_responses=True,
    compression_manager=compression_manager,
)

# ---------------------------------------------------------------------------
# Run Team
# ---------------------------------------------------------------------------
if __name__ == "__main__":
    research_team.print_response(
        "What are the latest developments in AI agents? Which companies dominate the market? Find the latest news and reports on the companies.",
        stream=True,
    )

Compression threshold

compress_tool_results_limit=2 triggers compression once at least two uncompressed tool-result messages are present. It does not preserve the latest two results uncompressed, despite the source comment. When triggered, the manager attempts to compress all eligible uncompressed tool results in the team's context.

The prompt requests preservation of facts and a 200-word limit; those are model instructions, not validated guarantees. Compare the compressed content with the original when evaluating the result.

Run the Example

Set up your virtual environment

uv venv --python 3.12
source .venv/bin/activate

Install dependencies

uv pip install -U agno ddgs openai sqlalchemy

Export your OpenAI API key

export OPENAI_API_KEY="your_openai_api_key_here"

Run the example

Save the code above as tool_call_compression_with_manager.py, then run:

python tool_call_compression_with_manager.py

Full source: cookbook/03_teams/10_context_compression/tool_call_compression_with_manager.py