The Agent Skills Directory

[Indirect Prompt Injection] (LOW): The skill identifies and reads external files provided as arguments, which can contain untrusted content.
Ingestion points: The skill reads files from paths like Quarto/ or Slides/ based on user-provided arguments.
Boundary markers: There are no explicit instructions or delimiters used to separate the slide content from the agent's instructions, nor are there warnings to ignore embedded commands.
Capability inventory: The skill uses Read, Write, and Task tools. While it primarily writes reports, the Task tool could be influenced by malicious instructions found within a slide file.
Sanitization: No sanitization or validation of the slide content is performed before it is processed by the analysis agents.
Risk: An attacker could embed hidden instructions in a LaTeX or Quarto file to manipulate the review scores or influence the agent's summary report.

slide-excellence