The Agent Skills Directory

[SAFE]: No malicious patterns, obfuscation, or unauthorized data exfiltration behaviors were detected. The skill correctly utilizes vendor-specific tools for its intended purpose of ad generation.
[PROMPT_INJECTION]: The skill exhibits an architectural surface for indirect prompt injection by ingesting external data (research briefs, competitor claims, and voice-of-customer quotes) from the advibly_get_brand_dossier tool (SKILL.md). This content is interpolated into generation prompts without boundary markers or sanitization. However, given the capabilities (social posting, video generation) are limited to the vendor's platform and the ingestion is required for the skill's primary purpose, the risk is assessed as safe. Evidence: 1. Ingestion points: advibly_get_brand_dossier; 2. Boundary markers: Absent; 3. Capability inventory: advibly_generate_video, advibly_social_create_post; 4. Sanitization: Absent.

ugc-ads