The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes creative content and metadata from third-party ad libraries, which creates a surface for indirect prompt injection attacks where malicious data could influence the agent's behavior.
Ingestion points: Competitor ad URLs and data exports from Meta Ad Library, TikTok Creative Center, Kalodata, and FastMoss.
Boundary markers: No explicit delimiters or isolation instructions are provided to separate untrusted external content from the skill's logic.
Capability inventory: The skill uses Bash(curl) for API communication, file system tools (Read/Write/Edit), and invokes secondary content generation skills.
Sanitization: Incorporates a mandatory ai-ad-compliance-check and an originality verification step to evaluate and filter generated concepts.
[COMMAND_EXECUTION]: The skill is configured to execute command-line utilities for data retrieval and media processing.
Evidence: Uses curl, jq, ffmpeg, and ffprobe as defined in the allowed-tools section to interact with the analysis API and manage video files.
[EXTERNAL_DOWNLOADS]: The skill communicates with external developer-managed infrastructure to perform core tasks.
Evidence: Fetches data from api.varg.ai for video analysis and references media stored on s3.varg.ai.

varg-ad-library-remix