The Agent Skills Directory

[SAFE]: The skill is purely instructional and contains no executable code or scripts.
[SAFE]: It mandates strict preservation of technical data including code blocks, environment variables, and API signatures, which prevents unintended modification of sensitive information.
[INDIRECT_PROMPT_INJECTION]: The skill processes external text for refinement, which is an ingestion surface for untrusted data. It mitigates this risk by instructing the agent to differentiate reference materials from the target text and lacks any permissions for high-risk tools or scripts in its configuration. Boundary markers are enforced through the instruction to separate draft and reference content.

fluent-english