shadcn-component-boundaries
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: The skill provides structural instructions and architectural guidelines for project file organization. It does not contain executable code, network requests, or attempts to harvest sensitive information.
- [METADATA_POISONING]: A discrepancy exists between the provided author context and the metadata author field in the skill file ('purrfold'). This appears to be a benign naming inconsistency rather than an attempt at malicious deception.
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to read
components.jsonto resolve path aliases. While this represents a data ingestion surface, the risk is minimal as it targets a standard local configuration file for path mapping. - Ingestion points:
components.json(SKILL.md Execution Step 1) - Boundary markers: None present
- Capability inventory: File system search and file write operations for component placement
- Sanitization: Not applicable for path resolution logic
Audit Metadata