Agno prompt compression

Integration

from agno.agent import Agent
from supercompress import Compressor

comp = Compressor()

agent = Agent(
    name="CompressedAgent",
    instructions="You are a helpful assistant.",
    show_tool_calls=True,
)

def run_compressed(agent, message):
    # Compress the context against the user message
    context = "\n".join(agent.memory.messages)
    result = comp.compress(context, message)
    agent.memory.messages = [result.compressed_text]
    return agent.run(message)

Frequently asked questions

Does it work with Agno's knowledge base?

Yes. Compress retrieved knowledge base content before passing it to the agent.

Can I use it with Agno's tool calls?

Yes. The compressor preserves tool descriptions and function schemas.