web-scraping-python

Warn

Audited by Gen Agent Trust Hub on Oct 22, 2026

Risk Level: MEDIUMPRIVILEGE_ESCALATIONDYNAMIC_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [PRIVILEGE_ESCALATION]: The scaffolding script modifies the file system permissions of generated files to make them executable by the user.
  • Evidence: Line 266 in scripts/new_scraper.py executes output_path.chmod(0o755) on the generated scraper script.
  • [DYNAMIC_EXECUTION]: The skill uses a template-based approach to generate executable Python scripts, which involves interpolating user-controlled strings into a code skeleton.
  • Evidence: In scripts/new_scraper.py, the SCRAPER_TEMPLATE string is populated using Template.safe_substitute with values derived from command-line arguments and subsequently written to disk using Path.write_text.
  • [INDIRECT_PROMPT_INJECTION]: The skill's 'Scraper Review' mode acts as a sink for untrusted data by instructing the agent to analyze and process code provided by external users.
  • Ingestion points: User-supplied scraper code snippets provided for review in SKILL.md (Mode 2).
  • Boundary markers: There are no specific instructions or delimiters provided to ensure the agent ignores instructions contained within the user-provided code.
  • Capability inventory: The agent has the ability to generate code, write files via provided scripts, and provide detailed technical advice.
  • Sanitization: The skill lacks mechanisms to sanitize or filter potential instructional content from the code being reviewed.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Oct 22, 2026, 02:29 PM