design-md-validator
Pass
Audited by Gen Agent Trust Hub on Aug 11, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes external DESIGN.md files and provides the output to the agent for triage. Malicious content within the design file's metadata or token descriptions could influence agent behavior.
- Ingestion points: The DESIGN.md file is read and parsed by the validate_design_md.mjs script.
- Boundary markers: The validator's findings are presented without strict delimiters to separate the tool's analysis from potentially malicious user-provided text in the design file.
- Capability inventory: The skill can execute the validation script and edit files in the workspace to fix identified issues.
- Sanitization: No sanitization is performed on the content of the DESIGN.md files prior to processing or interpreting results.
- [EXTERNAL_DOWNLOADS]: The validation script downloads the @google/design.md package via npx from the public NPM registry. This package is managed by a well-known organization.
- [REMOTE_CODE_EXECUTION]: The skill executes code downloaded at runtime from the NPM registry when running the design system validator CLI.
- [COMMAND_EXECUTION]: The script validate_design_md.mjs uses node:child_process.spawnSync to execute the npx command with the design linter. The arguments are passed as an array, minimizing typical shell injection vulnerabilities.
Audit Metadata