presenting-conference-talks
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill reads and processes research papers in PDF or LaTeX format to extract content for slides. This creates a vulnerability where an attacker could embed malicious instructions in a research paper that might influence the agent's behavior during the slide generation process.\n
- Ingestion points: Research paper files (PDF or LaTeX source) are ingested during the 'Content Extraction' phase described in
SKILL.md.\n - Boundary markers: There are no explicit delimiters or instructions provided to the agent to disregard potential commands or adversarial instructions found within the research paper text.\n
- Capability inventory: The skill involves executing shell commands such as
latexmkfor PDF generation andpython3for running the PPTX generation script provided inreferences/slide-templates.md.\n - Sanitization: No specific sanitization, filtering, or escaping logic is described for the content extracted from research papers before it is interpolated into slide templates or Python scripts.\n- [COMMAND_EXECUTION]: The skill instructs the agent to execute shell commands, specifically
latexmkto compile LaTeX Beamer slides andpython3to run a generated script for PPTX files. These are core functionalities for the skill's intended purpose and are safe when used with trusted papers.\n- [DYNAMIC_EXECUTION]: The skill uses a Python template to generate code that is subsequently executed to create PPTX slides. This involves assembling executable code based on paper metadata at runtime.
Audit Metadata