The Agent Skills Directory

[Prompt Injection] (SAFE): The documentation files provide templates and examples for output formatting and workflow structures. No instructions were found that attempt to override AI safety constraints or bypass system instructions.
[Data Exposure & Exfiltration] (SAFE): The scripts interact exclusively with the local filesystem to package skill directories. There are no hardcoded credentials, network requests, or attempts to access sensitive system paths (e.g., ~/.ssh, ~/.aws).
[Unverifiable Dependencies & Remote Code Execution] (SAFE): The scripts do not download or execute remote code. There is no usage of dangerous functions like os.system or subprocess.run to execute shell commands.
[Dynamic Execution] (SAFE): The quick_validate.py script uses yaml.safe_load() to parse frontmatter, which is a secure method that prevents code execution during deserialization. No usage of eval() or exec() was detected.
[Indirect Prompt Injection] (LOW): The skill validates external metadata from SKILL.md files. This is a potential attack surface, but the risk is mitigated by the use of safe YAML parsing and strict character validation (e.g., blocking angle brackets in descriptions).

skill-creator