Skip to content

Quick Start

This guide walks you through running your first AGENT-K mission using the API or programmatically.

Prerequisites

Make sure you have completed the installation steps:

  • Python 3.11+ with uv installed
  • uv sync completed in backend/
  • Environment variables configured (Kaggle + model API key)

Start the API Server

cd backend
python -m agent_k.ui.agui

The API server runs on http://localhost:9000.

Start a Mission (Chat Endpoint)

The /agentic_generative_ui/ endpoint accepts Vercel AI chat messages. When a mission intent is detected, it runs the mission and streams events.

curl -N -X POST http://localhost:9000/agentic_generative_ui/ \
  -H "Content-Type: application/json" \
  -d '{"id":"demo","messages":[{"role":"user","parts":[{"type":"text","text":"Find a Kaggle competition with a $10k prize"}]}]}'

Programmatic Usage

import asyncio
from agent_k import LycurgusOrchestrator
from agent_k.core.models import MissionCriteria

async def main():
    async with LycurgusOrchestrator() as orchestrator:
        result = await orchestrator.execute_mission(
            competition_id="titanic",
            criteria=MissionCriteria(
                target_leaderboard_percentile=0.10,
                max_evolution_rounds=50,
            ),
        )
        print(f"Final rank: {result.final_rank}")

asyncio.run(main())

Tooling Notes

  • web_search and web_fetch are built-in tools and only available for supported providers.
  • memory is only available for Anthropic models and stores files under .agent_k_memory by default.
  • Kaggle operations use the Kaggle adapter when credentials are available; otherwise OpenEvolve is used.

Using the Dashboard

For a visual interface, start both servers:

./run.sh

Then open http://localhost:3000 in your browser.

Next Steps