Sentinel

AI-assisted security and pentest terminal tool.

Overview

Sentinel is a modular terminal tool that combines an AI-powered assistant with structured security testing capabilities. It supports autonomous pentest campaigns with persistent state, multi-phase workflows, and evidence-backed reporting.

Quick Start

./sentinel                        # Interactive session
./sentinel --help                 # Show available options
./sentinel --pentest              # Start a pentest campaign session
./sentinel -p "your prompt"       # Single prompt mode

Key Features

• Provider system — Pluggable AI backend support (Z.ai, Anthropic-compatible)
• Pentest campaigns — Autonomous multi-phase security assessments
• Structured persistence — JSONL-based campaign state that survives interruptions
• Capability packs — Modular security testing coverage (web, API, cloud, network)
• Tool ecosystem — Built-in tools for file editing, code search, web fetching, and more

Pentest Mode

Launch a security assessment with natural language:

./sentinel --pentest
> Run a passive security audit on example.com

The AI operator will:

1. Initialize a campaign with scope and objectives
2. Advance through phases (recon, enumeration, testing, reporting)
3. Record operations, assets, observations, hypotheses, and findings
4. Generate a final report with evidence-backed findings

Campaign data is persisted under .sentinel-home/runs/.

Running Locally

./sentinel

Requires Bun runtime.

Tech Stack

Category	Technology
Runtime	Bun
Language	TypeScript (strict)
Terminal UI	React + Ink
CLI Parsing	Commander.js
Schema Validation	Zod v4
Protocols	MCP SDK, LSP

License

This project is for educational and security research purposes.