The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill requires the installation of the privacy-mask Python package from PyPI and the tesseract OCR engine through system package managers (Homebrew or APT). These are third-party dependencies maintained by the fullstackcrew-alpha community.
Evidence (SKILL.md): pip install privacy-mask, brew install tesseract, apt install tesseract-ocr.
[COMMAND_EXECUTION]: The skill interacts with the file system by executing the privacy-mask CLI tool. It interpolates user-provided file paths directly into shell commands, which could lead to command injection if the input paths are not strictly sanitized by the calling agent.
Evidence (SKILL.md): privacy-mask mask /path/to/screenshot.png, privacy-mask mask <path> --dry-run.
[PROMPT_INJECTION]: As the skill is designed to process external image data via OCR, there is a theoretical surface for indirect prompt injection if the text extracted from an image contains malicious instructions that the agent might follow in subsequent steps.
Ingestion points: Image files (PNG, JPG, BMP, etc.) processed by Tesseract/RapidOCR.
Boundary markers: The skill does not explicitly define prompt delimiters for the output of the masking tool.
Capability inventory: Shell command execution (privacy-mask CLI).
Sanitization: The tool's primary function is to redact sensitive strings, which serves as a security control, though it does not specifically filter for malicious instructions.

privacy-mask