design-handoff-reader
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from design handoff bundles, including project READMEs and chat histories. This content is analyzed and used to generate a report and direct instructions for the Claude Code agent. A malicious actor could embed hidden instructions within the bundle's text files to influence the agent's summary or manipulate downstream actions in the Claude Code environment.
- Ingestion points: Untrusted bundle content from ZIP files, local directories, or remote URLs.
- Boundary markers: None identified in the workflow to separate analyzed content from the skill's own processing instructions.
- Capability inventory: Generates text summaries and specific instruction strings for other agent environments.
- Sanitization: No explicit sanitization or filtering logic for external content is described.
- [EXTERNAL_DOWNLOADS]: The skill is designed to fetch and analyze content from external URLs provided by the user. This exposes the agent to potentially malicious remote content and dependencies on external server infrastructure during the analysis process.
Audit Metadata