The Agent Skills Directory

[SAFE]: No malicious patterns such as credential theft, data exfiltration, or unauthorized remote code execution were found. The skill operates within the expected scope of an image-processing assistant.\n- [SAFE]: The skill implements several security best practices, including explicit instructions to avoid sending research materials (screenshots, reference portraits) to external tools and strict adherence to user-provided white-lists for text content to prevent hallucinations or data invention.\n- [SAFE]: All external resources and licensing information correctly point to the author's official channels. No evidence of typosquatting or obfuscation was detected across the 65 files in the package.\n- [INDIRECT_PROMPT_INJECTION]: The skill possesses an attack surface as it processes untrusted user input (photos and titles) into tool prompts. However, this is mitigated by a mandatory evidence chain: 1) Ingestion point is the user input section in SKILL.md; 2) Boundary markers are enforced via the references/cover-prompt-blueprint.md template; 3) Capability is limited to built-in image tools; 4) Sanitization is performed through strict instructions to retain title verbatim and not invent external identity data.

punk-ip-video-cover