The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from an external source through the get_trends tool (Instagram trend data). This data is used to generate content briefs which can then be automatically published via the publish_post tool, creating a pathway for external content to influence agent actions.
Ingestion points: External data fetched via get_trends and user-provided brand profiles.
Boundary markers: No explicit delimiters or instructions are provided to the agent to treat external trend data as untrusted or to ignore embedded instructions.
Capability inventory: The skill possesses the ability to create and publish content to external platforms using create_post and publish_post tools.
Sanitization: There is no evidence of validation or sanitization of trend data before it is interpolated into content generation prompts.
[METADATA_POISONING]: The skill instructions and documentation contain assertive claims about a "2026 Instagram trust score update" and a "hard limit of 5 hashtags" that supposedly "blocks publication." These claims are currently inaccurate and may mislead the agent or the user regarding platform constraints and the necessity of specific warmup behaviors.
[EXTERNAL_DOWNLOADS]: The README.md instructs users to install the skill via bunx skills add genfeedai/skills/instagram-warmup. This involves remote code execution from a vendor-controlled repository. While this is a standard installation pattern for the Genfeed platform, it remains an external code reference.

instagram-warmup