The Agent Skills Directory

[SAFE]: The skill maintains a strong security posture by explicitly forbidding the upload of documents to external services and prohibiting the use of git or unauthorized network operations.
[INDIRECT_PROMPT_INJECTION]: The skill processes external document files (PDF, DOCX, PPTX, XLSX), which constitutes a potential ingestion surface for untrusted data. This is handled defensively within the skill's design.
Ingestion points: Office documents (PDF, DOCX, PPTX, XLSX) identified in Step 3 of the SKILL.md workflow.
Boundary markers: The 'Anti-Patterns' section in SKILL.md explicitly instructs the agent to avoid treating external content, logs, or tool responses as trusted instructions.
Capability inventory: The skill uses standard shell and editor tools for file inspection and local modification as noted in the SKILL.md workflow.
Sanitization: The primary purpose of the skill is to perform metadata sanitization, flagging and removing potentially sensitive information before files are shared.

document-metadata-review