The Agent Skills Directory

[SAFE]: The skill is focused on document translation and does not request excessive permissions, network access, or sensitive file operations.
[NO_CODE]: The skill consists entirely of instructional Markdown files and reference documents without any executable scripts or binary files.
[PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it is designed to ingest and process untrusted README files.
Ingestion points: The agent reads the contents of the source README.md and any existing localized siblings (as described in SKILL.md Workflow Steps 1 and 2).
Boundary markers: The instructions do not define explicit delimiters for untrusted content, but they provide a conceptual boundary by instructing the agent to audit the document structure separately from the prose.
Capability inventory: The skill involves file-writing operations to create or update localized README files (e.g., README.zh.md).
Sanitization: The instructions include extensive rules to avoid translating or modifying technical content such as code fences, shell commands, environment variables, and URLs, which serves to preserve the original technical intent of the document and limit the impact of potential injection in prose.

readme-i18n