director-master
Pass
Audited by Gen Agent Trust Hub on Sep 18, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external content, such as user-provided scripts, outlines, and text summaries, as defined in the input compatibility section of
SKILL.md. - Ingestion points: User scripts, rough text, and outlines are ingested in Step 1 and processed throughout the five-step workflow.
- Boundary markers: The skill body and reference files do not define explicit boundary markers or delimiters (like XML tags or triple quotes) to isolate user-provided text from the agent's internal instructions.
- Capability inventory: The skill performs complex analysis, text generation, and file writing (XLSX output in Step 5), which provides a surface for multi-step injection attacks where user input could attempt to manipulate the storyboard generation or file output logic.
- Sanitization: There are no explicit instructions for sanitizing, escaping, or validating the input data before processing it through the workflow steps.
- [DYNAMIC_EXECUTION]: Step 5 of the workflow in
SKILL.mdrequires the conversion of Markdown tables into a.xlsxfile. - Evidence: "Step 5: Output Table ... Convert Markdown storyboard to xlsx format".
- Details: This operation typically involves the agent generating and executing code (e.g., Python scripts using libraries like pandas or openpyxl) to create the binary file. This represents dynamic execution of logic derived from previously processed user content.
Audit Metadata