The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingest untrusted data from external LinkedIn profiles and previous post examples, creating an attack surface where malicious instructions could be embedded in the analyzed content.\n- Ingestion points: The skill requests LinkedIn profile links and post examples in the 'Discovery Questions' section of SKILL.md, and the 'Engagement Pattern Analysis' feature is designed to process historic engagement data.\n- Boundary markers: The instructions do not include specific delimiters or 'ignore' commands to prevent the agent from potentially executing instructions found within the LinkedIn content it is directed to analyze.\n- Capability inventory: The skill is restricted to text generation and content repurposing; no high-risk capabilities like file system access, network operations, or shell execution are requested or present in the skill body.\n- Sanitization: There is no evidence of filtering or validation logic intended to sanitize the external content before it is processed for storytelling analysis.

composing-linkedin-stories