The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands for directory management and document conversion.
mkdir -p ./prd_outputs/[Project Name]/ in Step 6a.
npx md-to-pdf "./prd_outputs/[Project Name]/[project_name]_PRD.md" in Step 6c.
[DYNAMIC_EXECUTION]: The shell commands in Step 6 interpolate the [Project Name] variable, which is derived directly from user input in Step 3. There are no sanitization steps or instructions to handle shell-sensitive characters (like semicolons, backticks, or quotes), creating a risk where a malicious user could provide a project name that executes arbitrary shell commands.
[EXTERNAL_DOWNLOADS]: The skill uses npx to fetch and run the md-to-pdf package from the public npm registry. While md-to-pdf is a known utility, fetching and executing code at runtime from an external registry introduces a dependency on external infrastructure.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted user input (product concepts) to generate a structured PRD that is intended to be fed into other AI coding tools.
Ingestion points: The user's initial product description and subsequent answers to up to 7 clarifying questions.
Boundary markers: No specific delimiters or warnings to ignore embedded instructions are provided in the generation logic.
Capability inventory: File system write access, directory creation (mkdir), and shell execution via npx.
Sanitization: No sanitization of user-provided strings is mentioned before they are interpolated into the generated markdown or shell commands.

prd-generator