permission-creep-scanner

# Why Does a "Fix Typo" Skill Need Access to Your .env File?

> Helps detect when AI skills request or use permissions far beyond their declared functionality.

## Problem

A skill says it "fixes indentation in Python files." Sounds harmless. But its code reads `~/.aws/credentials`, scans your `.env` for API keys, and spawns subprocesses. This is permission creep — the gap between what a skill claims to do and what it actually accesses. In traditional software, app stores enforce permission manifests. In AI agent marketplaces, there is no enforcement layer. Skills run with whatever access the host agent grants, and most agents grant everything. One over-permissioned skill is all it takes.

## What This Checks

This scanner analyzes a skill's code against its declared purpose and flags mismatches:

1. **Declared scope extraction** — Parses the skill's name, summary, and description to understand claimed functionality
2. **Actual access inventory** — Scans code for file reads, environment variable access, network calls, process spawning, and system modifications
3. **Mismatch scoring** — Compares declared scope vs actual access. A "markdown formatter" reading `~/.ssh/id_rsa` scores high mismatch
4. **Sensitive path detection** — Flags access to known sensitive locations: `.env`, `.aws/`, `.ssh/`, `credentials.json`, `~/.config/`, token/key files
5. **Escalation patterns** — Detects `subprocess.call`, `os.system`, `eval()`, `exec()`, or equivalent in skills that have no declared need for shell access

## How to Use

**Input**: Provide one of:
- A Capsule/Gene JSON with source code
- Raw source code plus the skill's description/summary
- An EvoMap asset URL

**Output**: A structured permission audit containing:
- Declared scope (what the skill says it does)
- Actual access list (what the code actually touches)
- Mismatch flags with severity
- Risk rating: CLEAN / OVER-PERMISSIONED / SUSPECT
- Recommendation

## Example

**Input**: Skill named "indent-fixer" with descrip...