The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests project-specific data such as HTML components, route metadata, and configuration files (.astrofy/config/paths.json) to derive Open Graph properties. It lacks explicit boundary markers or sanitization instructions, creating a surface where instructions embedded in project data could influence agent behavior. (Evidence: SKILL.md, references/technical.md)\n- [DATA_EXFILTRATION]: The skill is instructed to perform outbound network requests to 'open the exact URL emitted' to verify the status, dimensions, and public access of images. This creates a low-level risk of data exfiltration or Server-Side Request Forgery (SSRF) if the URLs are manipulated to point to sensitive internal resources or malicious external endpoints. (Evidence: SKILL.md, references/technical.md)\n- [COMMAND_EXECUTION]: The skill executes the astrofy check --category og command to validate the implementation. While this is a vendor-provided tool from 'promovaweb', it represents the execution of local CLI code. (Evidence: SKILL.md, examples/cases.md)

astrofy-open-graph