cpf-skill-creator
Pass
Audited by Gen Agent Trust Hub on Jun 27, 2026
Risk Level: SAFECOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill executes Python scripts from a specific local directory associated with official plugin installations (~/.claude/plugins/cache/claude-plugins-official/skill-creator/...). This is used for benchmarking, grading, and packaging tasks.\n- [COMMAND_EXECUTION]: The workflow interpolates user-controlled inputs, such as skill_name, directly into shell command templates (e.g., python -m scripts.aggregate_benchmark ... --skill-name ${inputs.skill_name}). This represents a potential command injection surface if the execution environment does not perform sanitization.\n- [PROMPT_INJECTION]: The skill processes untrusted user input (intent, feedback) to generate and modify SKILL.md files. It lacks explicit boundary markers or sanitization for this ingested data, making it a surface for indirect prompt injection, which is a common risk in automated code-generation tasks.
Audit Metadata