The Agent Skills Directory

[SAFE]: Extensive review of the skill's documentation, code patterns, and grading scripts revealed no malicious intent, data exfiltration, or unauthorized system access. The provided scripts perform static analysis of code artifacts for educational assessment purposes, avoiding dynamic execution of untrusted code.
[INDIRECT_PROMPT_INJECTION]: The skill's primary function includes reviewing and grading user-supplied code (e.g., domain.py), which is a known surface for indirect prompt injection. Malicious instructions could theoretically be placed in code comments or strings to influence the grading agent. 1. Ingestion points: graders/check.py reads domain.py and context-map.md from the local workspace. 2. Boundary markers: No specific boundary markers or instructions to ignore embedded content are used in the files processed. 3. Capability inventory: The agent has file system and command execution capabilities necessary for software development tasks. 4. Sanitization: The graders/check.py script utilizes the ast (Abstract Syntax Tree) module for static inspection. This ensures that the user-provided code is parsed as data and not executed as code, mitigating typical code execution risks.

ddd