The Agent Skills Directory

[SAFE]: The skill uses local Python scripts (init_crew.py, tools.py, copy_skill_to_workspace.py) to manage file operations and metadata. These scripts operate within the project scope and do not perform network requests or execute untrusted code.
[SAFE]: Data handling is restricted to the .workspace/ and .magic/ directories. Configuration data for initialization is stored in temporary JSON files before being processed into Markdown definitions, following standard management practices.
[SAFE]: The skill includes a 'Security Constraints' template in its reference documentation (prompt-engineering-guide.md), which advises users and agents to avoid revealing system prompts, performing destructive operations, or leaking sensitive info.
[SAFE]: Use of shell_exec is limited to executing known, local scripts included with the skill package for initialization and tool querying.

crew-creator