The Agent Skills Directory

[COMMAND_EXECUTION]: The script _postplus_shared/00-core/shared-runtime/scripts/download_videos_from_manifest_with_ytdlp.mjs executes the yt-dlp utility via spawn to download video content. This is a standard use of a well-known media processing tool.
[EXTERNAL_DOWNLOADS]: The skill fetches Instagram data and media files from remote URLs and the PostPlus Cloud API as part of its core research and collection workflows.
[INDIRECT_PROMPT_INJECTION]: The skill ingests and processes untrusted user-generated content from Instagram, representing an attack surface.
Ingestion points: Captions, comments, and profile biographies are ingested into JSON datasets processed by scripts like scripts/cluster_instagram_comments.mjs and scripts/rank_instagram_creators.mjs.
Boundary markers: Absent; the ingested text is processed directly as strings for classification and scoring purposes.
Capability inventory: File system access for dataset management, network operations for media retrieval, and subprocess execution for video processing.
Sanitization: Includes routine string cleaning and normalization (e.g., via cleanString in scripts/lib/instagram_common.mjs), though no specific sanitization for embedded prompt instructions is present.

instagram-tools