meccha-chameleon-game-trainer-detection
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill includes Python code snippets that utilize the
ctypeslibrary to interact with Windows system APIs (kernel32.dll), enabling process and memory manipulation such asOpenProcess,ReadProcessMemory, andWriteProcessMemoryfor analysis purposes. - [EXTERNAL_DOWNLOADS]: The instructions reference a sample ZIP archive hosted on Netlify, a well-known cloud hosting service, intended for research and detection analysis.
- [DATA_EXFILTRATION]: The skill provides patterns for communicating with external services, such as submitting files to the VirusTotal API for scanning and monitoring network connections using the
requestsandscapylibraries. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and analyze untrusted content from game trainer repositories, which presents a surface for indirect instructions.
- Ingestion points: External game trainer repositories, source code, and metadata (SKILL.md).
- Boundary markers: The instructions include a "Critical Security Notice" and explicit guidance for "Isolated Analysis Environment Setup" (e.g., Windows Sandbox, Virtual Machines).
- Capability inventory: The skill provides functionality for system API interaction, network communication, and deep file analysis using the
pefilelibrary (SKILL.md). - Sanitization: The provided code snippets do not describe specific sanitization or validation of the external content prior to analysis.
Audit Metadata