The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses local shell scripts (scripts/detect_i18n_profile.sh and scripts/validate_i18n_basics.sh) to perform static analysis of the repository. These scripts rely on standard tools like ripgrep (rg) to identify framework usage and localizable content within the current working directory.
[DATA_EXPOSURE_AND_EXFILTRATION]: No network activity or access to sensitive configuration files (e.g., SSH keys, cloud credentials) was observed. Data collected by scripts is saved locally as i18n-profile.json to facilitate the agent's workflow.
[INDIRECT_PROMPT_INJECTION]: As the skill involves reading and rewriting source code, it possesses an inherent surface for indirect prompt injection from repository content. This is a functional requirement for code-modifying agents and is not being exploited maliciously here.
Ingestion points: Reads source code in src, app, pages, and components directories via ripgrep.
Boundary markers: Relies on host agent implementation and platform-native file-edit tools.
Capability inventory: File read/write access via host platform and local script execution.
Sanitization: Results are processed into counts or structured JSON locally.

frontend-i18n-orchestrator