The Agent Skills Directory

[SAFE]: The skill implements a robust authorization pattern, using the AskUserQuestion tool to seek explicit user permission before performing any write operations to the file system (e.g., updating string tables, writing translator briefs, or saving QA reports).
[SAFE]: All file system interactions are scoped to relevant project directories such as src/, assets/, and production/localization/. No access to sensitive system paths or credentials was detected.
[SAFE]: The use of the Bash tool is limited to benign search operations using grep and glob patterns to identify hardcoded strings, audio references, and engine-specific configuration flags.
[INDIRECT_PROMPT_INJECTION]: The skill processes external project documentation and narrative files to generate translator briefings and cultural reviews. While this represents a data ingestion surface, the risk is mitigated as the skill uses this data for report generation rather than as input for executable commands or dynamic code generation.

localize