The Agent Skills Directory

[SAFE]: The skill is primarily educational, providing patterns and code examples for using native LLM structured output features and validation libraries like Pydantic and Zod.
[COMMAND_EXECUTION]: The skill includes scripts/verify.sh, which is a utility for linting the skill's own Markdown files. It uses grep and awk for local pattern matching and does not perform network requests or access sensitive system resources.
[INDIRECT_PROMPT_INJECTION]: The skill addresses the surface area for indirect prompt injection by teaching developers to use constrained decoding and robust validation layers. It provides specific patterns to prevent the model from fabricating data or ignoring schema constraints when processing untrusted inputs.
[REMOTE_CODE_EXECUTION]: The code examples utilize standard, well-known libraries such as instructor and pydantic for schema enforcement. No unverified dependencies or remote code execution patterns were found.

structured-extraction