The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external text files through its analysis scripts, which constitutes a potential attack surface for indirect prompt injection.
Ingestion points: The scripts scripts/brand_voice_analyzer.py and scripts/seo_optimizer.py ingest untrusted data by reading files passed as command-line arguments.
Boundary markers: There are no explicit boundary markers or instruction-ignoring warnings within the scripts to separate data from instructions.
Capability inventory: The skill possesses minimal capabilities; it performs read-only operations on text and outputs statistical analysis. No network operations, file-writing, or system-level execution based on file content were detected.
Sanitization: No sanitization of the input text is performed before processing, although the processing is limited to regex-based analysis.
[SAFE]: The Python scripts (brand_voice_analyzer.py and seo_optimizer.py) use only Python standard libraries (re, json, sys, typing) and do not perform any hidden network requests or unauthorized file access. The shell commands provided in the documentation (grep, cp, python) are standard, benign operations within the scope of content management.

content-creator