The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill references and encourages downloading installation resources and configuration from a remote repository at raw.githubusercontent.com/inference-sh/skills/.
[COMMAND_EXECUTION]: The skill instructs the user to execute commands using the belt CLI and npx, including belt login for authentication and belt app run to execute remote services like tavily/search-assistant and x/post-create.
[PROMPT_INJECTION]: The skill possesses an indirect prompt injection attack surface.
Ingestion points: External data is brought into the agent's context through the tavily/search-assistant tool which retrieves web content (SKILL.md).
Boundary markers: There are no explicit markers or instructions provided to the agent to treat the ingested data as untrusted or to ignore embedded instructions.
Capability inventory: The skill can perform external write operations via x/post-create and content generation through infsh/html-to-image (SKILL.md).
Sanitization: No validation or sanitization processes are described for the data retrieved from external tools before it is processed by other components.

linkedin-content