AGENTIX KDD

---

In one sentence

**Agentix KDD turns your repository's accumulated knowledge into an active prevention force: it makes your coding AI remember the project, not break what already worked, and leave a verifiable trail of every decision.**

It's not another AI that codes for you. It's the **armor** you put on the AI you already use — native on **Claude Code and Cursor** — and it lives **inside your project**: local SQLite, no cloud, no account, no subscription.

*KDD = Knowledge-Driven Development — development guided by the project's own accumulated knowledge. (npm package: `agentic-kdd`.)*

---

The problem it solves

You open Cursor or Claude Code. You explain your project *again*. The AI starts from zero *again*. It breaks something that was working *again*. It changes a business rule without remembering why it was set that way. Two real client cases motivated the current generation: a combobox applied "everywhere" broke selects that ALREADY worked, and CSS work broke existing `required` validations. Both are the same disease: **the AI can't see what's already proven, and nothing mechanical stops it.**

You're not coding — you're babysitting the context by hand. **Agentix takes that over.**

---

The full map — three pieces, and EVERYTHING hangs from one of them

Agentix has many organs but only three pieces. If you ever get lost in the feature list, come back here: **everything it does belongs to one of