The Agent Skills Directory

[SAFE]: The skill consists entirely of technical writing guidelines (incorporating Diátaxis, Google developer style, and Simplified Technical English) and does not contain any executable code, network operations, or sensitive data access.
[INDIRECT_PROMPT_INJECTION]: While the skill is designed to process untrusted data such as user-provided documentation, RFCs, and PR descriptions, it does not possess any dangerous capabilities such as file writing, command execution, or network access. The risk of indirect injection is mitigated by the skill's narrow instructional focus and the lack of systemic capabilities.
[METADATA_POISONING]: The skill's name, description, and internal rules are consistent with its stated purpose of improving technical writing quality. The disable-model-invocation: true flag in the metadata is a security best practice that prevents the agent from calling external tools while these instructions are being applied.

technical-writing