defuddle
Pass
Audited by Gen Agent Trust Hub on Oct 15, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill suggests installing the
defuddleCLI globally usingnpm install -g defuddle, which involves downloading software from an external registry. - [INDIRECT_PROMPT_INJECTION]: The skill processes content from arbitrary web pages, creating a risk that malicious instructions in the retrieved content could manipulate the agent.
- Ingestion points: Web content is parsed via the
defuddle parse <url>command. - Boundary markers: The instructions do not recommend using delimiters to isolate untrusted web content.
- Capability inventory: The skill uses a CLI parser; it does not expose dangerous capabilities like local file writes.
- Sanitization: The tool removes page clutter but does not provide security-specific sanitization against prompt injection.
Audit Metadata