implement-design
Pass
Audited by Gen Agent Trust Hub on Jun 19, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill uses the
WebFetchtool to acquire design source files fromapi.anthropic.com. This is a well-known service used for sharing Claude Design artifacts, and the operation is necessary for the skill's primary function of porting mockups to code. - [PROMPT_INJECTION]: The skill ingests untrusted data from external URLs or user-provided JSX/HTML, which presents a surface for indirect prompt injection. The risk is mitigated by the 'inventory-first' protocol, which forces the agent to map visual primitives to an allowed list of design system components before implementation.
- Ingestion points: External Claude Design URLs and user-pasted JSX (SKILL.md Step 0).
- Boundary markers: Absent; the skill does not use specific delimiters to separate untrusted data from the system prompt.
- Capability inventory: The skill has permissions to modify the project's codebase using
WriteandEdittools as specified in the frontmatter. - Sanitization: No explicit sanitization or filtering of the fetched/pasted content is performed.
Audit Metadata