The Agent Skills Directory

[COMMAND_EXECUTION]: The scripts/check.py utility uses subprocess.run to execute local Python scripts (validate_skill.py, validate_evals.py, package.py, audit.py) within the skill's own directory structure. These commands are part of the validation workflow defined in the skill's operator contract to ensure integrity before deployment.
[PROMPT_INJECTION]: The skill processes content from justfile, Makefile, and package.json files, which constitutes a surface for indirect prompt injection. This risk is mitigated through instructions that enforce a write-scoped posture and require explicit user confirmation before executing side-effecting recipes or deleting source files.
[SAFE]: The skill implements proactive security measures, including explicit rules against hardcoding tokens and secrets in task runners. Additionally, the scripts/asset_toolkit/package.py script contains comprehensive safety logic to prevent directory traversal, symlink attacks, and the inclusion of non-portable absolute paths in packaged skills.

justfile