interview-cheatsheet

Pass

Audited by Gen Agent Trust Hub on Nov 10, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [SAFE]: The skill incorporates a robust security check aimed at identifying and redacting sensitive data, such as private server paths, degree affiliations, or specific laboratory names, before content is finalized.
  • [SAFE]: The workflow enforces human-in-the-loop security by prohibiting the agent from automatically committing or pushing files, ensuring all generated documentation undergoes user review.
  • [INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted user input in the form of a tutorial topic, which is then used to generate a large markdown file subsequently parsed by a reviewer agent. * Ingestion points: User-provided topic argument and the generated draft file at docs/tutorials/_tutorial.md. * Boundary markers: The review agent is invoked with a fresh thread and a specific JSON-formatted task instruction to minimize accidental obedience to instructions within the tutorial text. * Capability inventory: The skill uses Read, Write, and Edit tools to manage local files and Bash to execute a rendering script. * Sanitization: Content is audited by a secondary high-reasoning agent specifically configured to detect errors and style violations.
  • [COMMAND_EXECUTION]: The skill uses the Bash tool to run a local Python rendering utility. This execution is limited to repository-resident scripts (skills/render-html/scripts/render_html.py) and is used to convert markdown tutorials into HTML format.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 10, 2026, 09:37 AM