The Agent Skills Directory

[Workspace File Access]: The skill provides instructions for reading test files using standard tools, with fallback logic for shell utilities like 'cat', 'sed', or 'Get-Content' to handle path normalization issues. These instructions explicitly emphasize the need to verify that file paths remain within the workspace boundaries to prevent unauthorized access.
[Indirect Prompt Injection Surface]: As a tool designed to analyze user-provided code, the skill has a surface for indirect prompt injection. A malicious test file could contain instructions intended to influence the agent's behavior. Evidence Chain: (1) Ingestion: Workspace test files identified in workflow Step 1/2. (2) Boundary markers: Not explicitly defined for embedded code instructions. (3) Capability: File-read operations and diagnostic reporting. (4) Sanitization: Not explicitly defined. The risk is considered low given the diagnostic intent of the skill.
[Modular Extension Design]: The skill attempts to load optional language-specific extensions to enhance its analysis, falling back to its internal anti-pattern catalog if they are unavailable, which follows common extensible agent design patterns.

test-anti-patterns