frontend-css-strategy
Pass
Audited by Gen Agent Trust Hub on Sep 13, 2026
Risk Level: SAFEMETADATA_POISONINGINDIRECT_PROMPT_INJECTION
Full Analysis
- [METADATA_POISONING]: The SKILL.md file contains a large volume of documentation (lines 359-490) that is entirely unrelated to its stated purpose of 'frontend-css-strategy'. This section includes implementation patterns for event buses, deployment checklists for database migrations, monitoring metrics for disk/memory usage, and secrets management policies. Including high-privilege operations like credential rotation and secrets access auditing in a CSS strategy skill is deceptive and could mislead an agent into performing unauthorized or out-of-scope backend tasks.
- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided context such as framework details, team composition, and current styling approaches to generate code recommendations and file structures. The instructions explicitly direct the agent to create folder structures and configure build tools based on this input. The lack of strict boundary markers or sanitization requirements for this untrusted user context creates a surface for indirect prompt injection, which could lead to the generation of malicious configurations or shell-executable artifacts.
Audit Metadata