@4meta5/pi-ozcar
Structured Security Audit Extension Pack for the Pi Ecosystem
Install & Use
Copy this prompt and send it to your AI assistant (Claude / Cursor / TRAE / Codex / WorkBuddy etc.) to auto-install:
Help me install this AI Skill: @4meta5/pi-ozcar. It is used for: Structured Security Audit Extension Pack for the Pi Ecosystem Full Skill content: https://321skill.com/skills/4meta5-pi-ozcar/raw/index.md Read that page and install it.
The prompt includes a link to the full Skill content. You can also view the full content.
This Skill addresses the lack of a unified process and comparable, exportable results when conducting structured security reviews in AI-assisted development. In practice, security audits often rely on manual expertise, with audit records scattered across conversations, making them difficult to reproduce, compare, and archive.
Usage is straightforward. Simply install @4meta5/pi-ozcar in your Pi environment, then enter audit mode via the /ozcar command. From there, you can use natural language instructions to initiate an audit (/ozcar-audit-start <focus>), record hypothesis branches (/ozcar-audit-branch hypothesis <slug>), create checkpoint snapshots (/ozcar-audit-checkpoint <snapshot.json>), and export a structured JSON report (/ozcar-audit-export). The entire audit state is saved on Pi's /tree branch, supporting resumption after interruption.
It is well-suited for teams or individuals requiring a standardized security audit process, especially developers already using Pi as their AI Agent engine. Whether for smart contract audits, code security reviews, or compliance checks, this command system delivers reproducible and comparable audit results.
It is recommended to use this tool to initialize an audit session at the start of each security review. Note that it primarily relies on Pi's conversation and branching mechanisms, so familiarity with basic Pi operations is required. Also, the quality of audit results depends on the underlying model's capabilities; using it with a strong reasoning model is advised.
Key Features
Unlike conducting ad-hoc verbal audits directly in conversation, `pi-ozcar` provides full audit lifecycle management: branching hypotheses, checkpoint snapshots, and deterministic JSON exports. This makes the audit process reproducible and comparable, rather than just a collection of transient chat logs.
Limitations
Currently only supports the Pi Mono engine (`@badlogic/pi-mono`), requires a Node.js environment, and the audit export format is a fixed JSON schema, with no support for custom output templates.
FAQ
How do I install pi-ozcar?
Run `pi install npm:@4meta5/pi-ozcar` in your Pi environment for a persistent installation, or use `pi -e npm:@4meta5/pi-ozcar` to load it temporarily.
Where is the audit export file located?
After executing `/ozcar-audit-export`, structured files such as `exports/findings.json` are generated in the `.ai-auditor/` directory.
Installation guide for AI assistants
If your AI coding assistant (Claude Code, Cursor, TRAE etc.) can see this page, send it this message to auto-install:
Visit https://321skill.com/skills/4meta5-pi-ozcar/raw/index.md to read the original Skill definition (Markdown format) for @4meta5/pi-ozcar, and install it according to the instructions.
Raw Markdown URL for AI: /skills/4meta5-pi-ozcar/raw/index.md