The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes git diff via a Python script (scripts/find_new_translations.py) to identify changed translation files. This is a standard operation for determining local changes in a development environment. The subprocess.run calls use hardcoded command arguments (['git', 'diff', '--name-only']), preventing command injection.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted external data from XLF translation files and source code context (Step 1 and Step 2).
Ingestion points: Reads target and source text from XLF files via scripts/find_new_translations.py and reads source code via the Grep and Read tools.
Boundary markers: The skill uses clear visual delimiters (separator lines) when presenting translations for review in Step 3.
Capability inventory: The skill has the capability to write to the local filesystem (Step 4) and run local shell commands (Step 1).
Sanitization: There is no explicit sanitization of the XLF content before it is displayed to the user; however, the skill explicitly requires user confirmation ('Wait for user confirmation before proceeding') for every translation proposed, which mitigates the risk of an automated attack through malicious data ingestion.

translate