The Agent Skills Directory

[SAFE]: The skill is entirely defensive, providing guidelines, detection logic, and verification scripts to help users identify malicious Minecraft client repositories. It contains no hidden code, exfiltration logic, or unauthorized system access.
[INDIRECT_PROMPT_INJECTION]: The skill provides functions designed to process untrusted data from external repositories (metadata and README text) for malware analysis.
Ingestion points: analyze_minecraft_repository and analyze_readme_content functions in SKILL.md take external strings/dictionaries as input.
Boundary markers: Not implemented in the provided logic snippets.
Capability inventory: The analysis logic is restricted to regex matching, string counting, and local file hashing (hashlib). There are no network, file-write, or shell-execution capabilities associated with these analysis functions.
Sanitization: Not applicable as the data is used for pattern matching, not for command construction or prompt interpolation.
Severity is LOW because the attack surface is limited to analytical pattern matching without dangerous side effects.

minecraft-vape-v4-malware-detection