defuddle

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADSCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it fetches and processes content from external, untrusted web pages.
  • Ingestion points: The skill parses content from arbitrary user-provided URLs using defuddle parse <url> in SKILL.md.
  • Boundary markers: Absent. There are no instructions or delimiters defined to prevent the agent from obeying instructions embedded in the fetched web content.
  • Capability inventory: The skill utilizes subprocess execution to run the defuddle CLI tool and read its output.
  • Sanitization: Absent. Content is passed directly from the external source to the agent's context without filtering.
  • [EXTERNAL_DOWNLOADS]: The skill documentation encourages the installation of an external package from a public registry.
  • Evidence: npm install -g defuddle in SKILL.md.
  • [COMMAND_EXECUTION]: The skill's primary functionality relies on executing a CLI tool with user-provided arguments.
  • Evidence: Commands like defuddle parse <url> --md involve executing a shell command where the URL is a variable parameter.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 06:53 AM