The Agent Skills Directory

[COMMAND_EXECUTION]: The skill requires permission to execute the infsh command-line tool via Bash using a wildcard pattern (infsh *). This is used to authenticate the user and run remote AI models.
[EXTERNAL_DOWNLOADS]: The documentation references the installation of additional skill components using the npx skills add command, which downloads skill definitions from the inference-sh vendor repository.
[PROMPT_INJECTION]: The skill accepts user-provided text prompts and image URLs which are passed as arguments to the infsh command.
Ingestion points: User-controlled data enters the tool context via the --input flag in SKILL.md.
Boundary markers: Prompts are delimited within a JSON-formatted string passed to the CLI.
Capability inventory: The skill has the capability to execute system commands via the Bash tool as defined in the allowed-tools section of SKILL.md.
Sanitization: No explicit client-side sanitization is described in the skill instructions; it relies on the vendor's platform to handle inputs securely.

p-image