Skip to content

Latest commit

 

History

History

examples

README.md

Cortex Examples

Runnable examples demonstrating Cortex capabilities, from basic mapping to multi-agent framework integration.

Prerequisites

# Install the Cortex runtime
cargo install cortex-runtime

# Install the Python client
pip install cortex-agent

Cortex auto-starts the daemon and handles Chromium installation on first run. No manual setup needed.

Examples

# File Description Dependencies
01 01_quickstart.py Map a site and list pages in 10 lines cortex-agent
02 02_price_comparison.py Compare prices across 5 e-commerce sites cortex-agent
03 03_pathfinding.py Find shortest navigation route through a site cortex-agent
04 04_cross_site_comparison.py Compare structures of multiple websites cortex-agent
05 05_perceive_page.py Analyze a single page without full site mapping cortex-agent
06 06_langchain_agent.py Use Cortex tools in a LangChain agent cortex-langchain, langchain
07 07_crewai_researcher.py Build a CrewAI research crew with Cortex cortex-crewai, crewai
08 08_autogen_agent.py Register Cortex functions with AutoGen agents cortex-autogen, pyautogen
09 09_semantic_kernel.py Use Cortex as a Semantic Kernel plugin cortex-semantic-kernel, semantic-kernel
10 10_mcp_server_demo.py Simulate MCP tool calls (Claude, Cursor, etc.) cortex-agent
11 11_runtime_status.py Monitor runtime health and cached maps cortex-agent

Running

# Run any example
python examples/01_quickstart.py

# Framework examples work in two modes:
# 1. Direct mode (no LLM needed) — runs out of the box
# 2. Agent mode (uncomment LLM sections) — requires API keys

Framework Integration Quick Reference

LangChain

from cortex_langchain import CortexMapTool, CortexQueryTool
tools = [CortexMapTool(), CortexQueryTool()]
agent = initialize_agent(tools, llm, agent="zero-shot-react-description")

CrewAI

from cortex_crewai import CortexWebCartographer
agent = Agent(role="Researcher", tools=[CortexWebCartographer()])

AutoGen

from cortex_autogen import cortex_map, cortex_query
assistant.register_for_llm(name="cortex_map")(cortex_map)

Semantic Kernel

from cortex_semantic_kernel import CortexPlugin
kernel.add_plugin(CortexPlugin(), plugin_name="cortex")

MCP (Claude, Cursor, Windsurf, Continue)

cortex plug  # One command — auto-discovers and configures all agents

License

Apache-2.0