craft-sync
Pass
Audited by Gen Agent Trust Hub on Nov 15, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill is designed to execute local development commands as part of the export workflow. Specifically, it is instructed to run a project's verification checks (such as lint, typecheck, and tests) and utilizes a headless Chrome command to generate visual snapshots of HTML-based design guides.
- [INDIRECT_PROMPT_INJECTION]: The skill presents a surface for indirect prompt injection because it ingests and processes untrusted content directly from the codebase (e.g., token names, component variants, and documentation files) to influence the agent's actions and output.
- Ingestion points: Reads project files such as
tokens.css,tailwind.config.*, JSON token sources, UI component files (.tsx,.jsx), andDESIGN.md. - Boundary markers: The instructions do not specify a requirement for delimiters or system prompts that instruct the agent to ignore instructions embedded within the design data it processes.
- Capability inventory: The skill possesses the ability to write to project source files, update metadata (
design-sync.json), and execute system commands for project validation. - Sanitization: There is no evidence of sanitization or validation logic applied to the design token names or values parsed from the source code before they are used to generate exports or update the design tool.
Audit Metadata