agent-skill-builder
Pass
Audited by Gen Agent Trust Hub on Oct 8, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The skill provides instructions for the agent to execute shell commands (e.g.,
mkdir -p ~/.claude/skills/[skill-name]) to set up the necessary directory structure for new skills. This is a standard and expected operation for a scaffolding tool. - [INDIRECT_PROMPT_INJECTION]: The skill is designed to process instructions and knowledge files from external ChatGPT projects during migration. This ingestion of untrusted data represents a potential attack surface, though the skill's instructions focus on structural conversion and mapping rather than direct execution of the content.
- [DYNAMIC_CONTEXT_INJECTION]: The documentation references the platform's dynamic injection syntax (e.g.,
!command). The skill correctly identifies this as a platform feature and provides benign examples for development workflows, such as fetching git logs or pull request diffs, without using the syntax for malicious purposes.
Audit Metadata