The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute local Python scripts and standard shell commands (ls, grep, mkdir) as part of its documented workflow for skill discovery, creation, and validation. These operations are limited to the local file system within designated skill directories.
[PROMPT_INJECTION]: The skill possesses an ingestion surface for indirect prompt injection because it processes markdown files from the local file system.
Ingestion points: The discover_skills.py and validate_skill.py scripts read SKILL.md files from the local skills directory.
Boundary markers: Absent. The scripts parse YAML frontmatter but do not wrap the content in protective delimiters before presentation to the agent.
Capability inventory: The skill allows the execution of shell commands and local Python scripts, which are its primary tools for management.
Sanitization: Absent. Content read from skill files is processed without validation or escaping, although the risk is mitigated by the skill's specific developer-focused use case.
[SAFE]: No hardcoded credentials, sensitive file access (e.g., SSH/AWS keys), or remote code execution from untrusted network sources were identified. All included scripts are part of the skill's own package and perform legitimate management tasks.

skill-creator