The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill-report.json file contains a security_audit section with self-referential safety claims, providing a pre-computed "safe" verdict and technical summary asserting "zero attack surface."
Ingestion points: skill-report.json metadata file.
Boundary markers: None.
Capability inventory: The skill is a documentation-only resource; no executable scripts, network tools, or file system writing capabilities are included in the provided files.
Sanitization: Not applicable as the skill does not process external data or execute commands.
[COMMAND_EXECUTION]: The documentation in SKILL.md and references/skill-creator-original.md includes shell command examples for directory creation (mkdir), skill initialization (init_skill.py), and plugin testing (cc). These are presented as instructional examples for manual developer use and are not executed by the agent automatically.

skill-development