The Agent Skills Directory

[Prompt Injection] (HIGH): The skill is designed for 'Generative Engine Optimization' (GEO), which inherently involves modifying data to influence AI agent reasoning and citation logic. This creates a high-severity Indirect Prompt Injection vulnerability surface.
Ingestion points: scripts/auto_implementer.py accepts and processes arbitrary local HTML files via the file_path argument.
Boundary markers: There are no delimiters or 'ignore' instructions implemented to prevent the optimizer from being influenced by malicious instructions embedded in the source HTML.
Capability inventory: The skill can execute multiple subprocesses and write various new files to the local filesystem (e.g., -optimized.html, -voice.html).
Sanitization: Only basic regex-based HTML cleaning is present in scripts/shared/text_utils.py, which is insufficient to neutralize instructional injection or structural manipulation.
[Command Execution] (MEDIUM): scripts/auto_implementer.py orchestrates a multi-stage pipeline using subprocess.run. While it uses a safer list-based argument format, it attempts to execute six external Python scripts (audit_report.py, content_optimizer.py, platform_optimizer.py, voice_optimizer.py, citation_enhancer.py, and freshness_monitor.py) that are not included in the provided source code, rendering the final execution logic unverifiable and potentially unsafe if these scripts have been tampered with or contain vulnerabilities.

seo-geo-optimizer