The Agent Skills Directory

[SAFE]: The skill is a pure instructional router designed to guide the user to other specialized skills. It includes logic to check for the presence of a configuration file (social-context.md) to verify initial setup.
[DATA_EXPOSURE]: The skill reads from a local configuration file (social-context.md) to determine the user's setup status. This is a standard practice for managing agent state and does not involve network activity or external exfiltration.
[PROMPT_INJECTION]: No patterns indicative of prompt injection, jailbreaking, or safety filter bypasses were found. The instructions are clearly scoped to routing user requests to a fixed set of sub-skills.
[DYNAMIC_EXECUTION]: The skill uses static mapping and does not generate or execute dynamic code. The frontmatter includes disable-model-invocation: true, which further limits the execution environment to providing textual guidance.

idk