The Agent Skills Directory

Indirect Prompt Injection (HIGH): The skill creates a dangerous injection surface by ingesting untrusted data from arbitrary website URLs provided by users via the references/brand-extraction.js script. This data is then used to drive file generation in scripts/scaffold_deck.js and passed to the potxkit tool. There are no boundary markers or sanitization processes described to prevent malicious instructions embedded in the scraped websites from being obeyed by the agent.
Unverifiable Dependencies (MEDIUM): The skill relies on the pptxgenjs library and references an unverified third-party tool named potxkit (accessed via CLI or MCP), which is not within the trusted external source scope.
Command Execution (LOW): The workflow requires the agent or user to execute local Node.js scripts (scripts/scaffold_deck.js) to generate outputs, which is an expected but notable local execution surface.

brand-pptx-template