defuddle
Pass
Audited by Gen Agent Trust Hub on Sep 19, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is designed to fetch content from external URLs and provide it to the agent for analysis or summarization. This creates a surface for indirect prompt injection, where a malicious website could include hidden instructions to influence or hijack the agent's behavior once the content is processed.
- Ingestion points: The output of the
defuddle parse <url> --mdcommand is directly provided to the agent. - Boundary markers: There are no instructions to wrap the fetched content in delimiters or to ignore potential instructions embedded in the web data.
- Capability inventory: The agent uses the extracted content for analysis, summarization, and document processing.
- Sanitization: No methods for sanitizing or filtering the markdown output are mentioned or enforced.
- [EXTERNAL_DOWNLOADS]: The skill requires the installation of an external Node.js package (
defuddle) from a public registry that is not part of the standard environment. - Evidence: The documentation instructs the user/agent to run
npm install -g defuddleif the tool is missing.
Audit Metadata