vue-testing-best-practices
Pass
Audited by Gen Agent Trust Hub on Sep 30, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONMETADATA_POISONING
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill instructs the agent to analyze project-specific components and execute testing commands like
npx vitestandnpx playwright test. This creates a vulnerability surface where malicious code within an analyzed project could potentially influence the agent's behavior. \n - Ingestion points: Reactively reads and analyzes Vue components, composables, and stores. \n
- Boundary markers: Missing specific instructions to treat project code as untrusted data. \n
- Capability inventory: Recommends executing shell commands for package management and test running. \n
- Sanitization: No specific sanitization methods are provided for interpolating project code into prompts. \n- [METADATA_POISONING]: The skill's internal metadata identifies the author as
github.com/vuejs-ai, which differs from the platform-reported authorliangmiqwq. This discrepancy, while likely due to mirroring, could be misleading regarding the skill's provenance.
Audit Metadata