The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user text through the $ARGUMENTS variable to perform brand-voice transformations. While the skill lacks explicit boundary markers for this input, the potential impact is mitigated by the highly restricted set of allowed tools.
Ingestion points: User-supplied source text provided via command-line arguments in SKILL.md.
Boundary markers: Absent; the instructions do not explicitly wrap the input in delimiters or warn the agent to ignore instructions embedded within the user text.
Capability inventory: The skill is restricted to Read, Glob, and Grep on local files, and AskUserQuestion. It does not have network access or file-writing capabilities, preventing typical exfiltration or persistence vectors.
Sanitization: No explicit sanitization or filtering of the user input is performed before processing.

brand-quill