The Agent Skills Directory

[SAFE] (SAFE): The skill is entirely focused on defensive programming techniques for LLMs. It demonstrates how to use programmatic assertions to validate AI outputs against desired formats and content policies.
[EXTERNAL_DOWNLOADS] (SAFE): The code snippets reference standard and reputable libraries including dspy and pydantic. There are no commands to download or execute untrusted scripts or binary files.
[DATA_EXFILTRATION] (SAFE): No sensitive file access (e.g., SSH keys, credentials) or unauthorized network communication was detected. The examples use standard API configuration for LLM providers.
[COMMAND_EXECUTION] (SAFE): The skill does not utilize any dangerous functions such as eval(), exec(), or subprocess calls to execute shell commands.
[PROMPT_INJECTION] (SAFE): Rather than attempting to bypass safety filters, this skill teaches users how to implement stricter controls to prevent the model from deviating from its intended behavior.

ai-following-rules