The Agent Skills Directory

[COMMAND_EXECUTION]: The skill frequently executes external command-line utilities to perform its tasks, including git for version control, gh for GitHub interactions, and prettier for code formatting.
Evidence: Subprocess calls in scripts/translation_deploy.py and scripts/_support.py are implemented using argument lists, which is a secure practice to prevent shell injection.
[EXTERNAL_DOWNLOADS]: The skill downloads translation data from Memsource, an established Translation Management System (TMS).
Evidence: The download command in scripts/translation_deploy.py calls the translations-cli to fetch completed translation jobs from the Memsource API.
[INDIRECT_PROMPT_INJECTION]: The skill creates a surface for indirect prompt injection by ingesting external data from a TMS and writing it into the project's source code and configuration files.
Ingestion points: JSON files downloaded from the external TMS in scripts/translation_deploy.py.
Boundary markers: The script uses the _format_ts_string function to escape backslashes and quotes when merging translations into TypeScript files, reducing the risk of code injection.
Capability inventory: The skill has the capability to write to the local filesystem and execute commands via git and gh.
Sanitization: Basic JSON structure validation is performed in the _validate_translation_file function to ensure ingested data meets expected formats.
[DYNAMIC_EXECUTION]: The main Python script dynamically modifies the system path to facilitate the import of local support modules.
Evidence: Found sys.path.insert(0, str(_scripts_dir)) in scripts/translation_deploy.py, which is a common and low-risk pattern for managing local script dependencies.

rhdh-translation-deploy