# Project Overview — Agent-Kit (V3.1)

## Mission

To provide an elite IDE Augmentation Arsenal (via MCP) that injects multi-agent reasoning, specialized personas, and deterministic validation scripts into existing AI environments like Cursor and Copilot.

## Current State

- **Phase**: V3.1 Hybrid Arsenal Pivot Complete.
- **Hub Version**: 3.1.0
- **Primary Logic**: The Gemini 3.1 Cognitive Baseline (2M Ultra Context, ToT Architecture).

## Strategic Objectives

1.  **Invisible Integration**: Developers do not leave their IDE. We extend their IDE's capabilities natively via the Model Context Protocol.
2.  **Anti-Fragile Validation**: Expose Python security and codebase audits (`run_project_audit`, `run_security_chaos`) to the IDE AI to enforce stability.
3.  **Scale-Ready Intelligence**: 53 pre-trained agents injected with 2026/2027 enterprise-level reasoning models.