The Agent Skills Directory

[SAFE]: Comprehensive review of the skill's instructions and reference files revealed no malicious patterns, hardcoded secrets, or suspicious network activities. The skill's behavior is consistent with its stated purpose of auditing mobile app listings.- [NO_CODE]: The skill is entirely descriptive, containing no code, scripts, or binaries. This architectural simplicity eliminates vectors related to remote code execution or dynamic library injection.- [PROMPT_INJECTION]: The skill's functionality requires processing untrusted external content from app store listing pages, creating a surface for indirect prompt injection. While an attacker could attempt to manipulate the audit's findings through specially crafted app descriptions, this is a known limitation of web-analysis tools. 1. Ingestion points: External URLs (apps.apple.com, play.google.com) in SKILL.md. 2. Boundary markers: None specified. 3. Capability inventory: Use of WebFetch, screenshot capture, and local file reading (.agents/product-marketing-context.md). 4. Sanitization: No sanitization of ingested content is mentioned.

aso-audit