presenting-conference-talks
Pass
Audited by Gen Agent Trust Hub on Oct 1, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes research papers in PDF or LaTeX format, which are untrusted external sources that could contain embedded instructions intended to influence the agent's behavior during slide generation.
- Ingestion points: Workflow Step 1 in
SKILL.mdinstructions the agent to read the compiled paper source. - Boundary markers: None provided; the skill does not specify delimiters or instructions to ignore embedded commands within the paper content.
- Capability inventory: The skill includes file-writing capabilities (
prs.save) and shell command execution (latexmk,python3) viaSKILL.mdandreferences/slide-templates.md. - Sanitization: There is no evidence of sanitization or filtering of the input paper content before it is interpolated into slide generation logic.
- [DYNAMIC_EXECUTION]: The skill provides a Python script template and instructs the agent to write and execute it locally to generate PPTX files.
- Evidence:
SKILL.md(Workflow Step 4) andreferences/slide-templates.mdprovide a complete script (generate_slides.py) and execution instructions (python3 generate_slides.py). This is a standard automation pattern for this skill's purpose.
Audit Metadata