A hand-written agentic loop on the raw Anthropic SDK, wrapped in a Chainlit UI - a security-audit copilot for headers, TLS, CVEs, and DNS.

A sample agent project demonstrating how to build a tool-using AI agent from scratch, directly on the Anthropic API — no LangChain, no LangGraph, no CrewAI. It's themed as a security-audit "swiss army knife" chatbot, wrapped in a Chainlit UI, but the real point of this repo is the agentic loop architecture, not the security tooling itself.

If you're learning how to build agents with the Anthropic API, this is meant to be a readable, working reference: a hand-written loop under 100 lines, tools defined as plain Python functions, and a UI layer that's kept fully separate from the agent logic.....