The Agent Skills Directory

[SAFE]: The skill consists primarily of documentation and code snippets for API development. All code examples provided are static templates for REST and GraphQL design and do not contain any instructions that would lead to unauthorized execution or data access.\n- [DATA_EXPOSURE_AND_EXFILTRATION]: The SKILL.md file contains a section 'Emit Outcome Sidecar' which instructs the agent to write session metadata to a local JSON file in a specific directory (~/.claude/skill-analytics/). This is a legitimate logging practice for tracking skill usage and performance within a development environment and does not involve exfiltrating sensitive data to external services.\n- [EXTERNAL_DOWNLOADS]: The documentation references several well-known and trusted external tools such as Redocly, Scalar, and OpenAPI-typescript. These are standard developer utilities, and their inclusion in the guidance is for user implementation purposes rather than for immediate, automated, or risky remote execution.

api-design