The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input regarding the recipient and required information to generate a markdown file. This creates an attack surface where instructions could be embedded in the data if processed by another agent later, although the skill itself is passive.
Ingestion points: User responses captured during the interview steps in SKILL.md.
Boundary markers: The template uses Markdown structure (headers and blockquotes) but does not include explicit delimiters to prevent instruction injection.
Capability inventory: The skill writes a text file to the current directory using the agent's filesystem capabilities.
Sanitization: No specific filtering or escaping of user input is described before it is written to the output file.
[COMMAND_EXECUTION]: The skill instructs the agent to write a file (to-questionnaire-<slug>.md) to the local directory. This is the primary intended function of the skill and does not involve executing arbitrary shell commands or external scripts.

to-questionnaire