The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes a Python script, scripts/translation-drift-check.py, which executes shell commands via subprocess.run to interact with Git history. This script is used to calculate how many commits a translated page is behind its source. The script uses list-based execution (avoiding shell=True) and strictly validates commit hashes extracted from documentation files against a hexadecimal regular expression (HASH_RE), significantly mitigating the risk of command injection.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to process external documentation files and Git history to manage localization strategies. Maliciously crafted documentation files or commit metadata (e.g., from an external contributor's Pull Request) could attempt to influence the agent's recommendations or manipulate the drift detection output.
Ingestion points: The translation-drift-check.py script reads documentation files (.md, .mdx, .rst, .adoc) and frontmatter content. The agent also ingests user-provided documentation statistics during the interview phase.
Boundary markers: The skill instructions do not explicitly require the agent to use delimiters or specific safety instructions when processing untrusted documentation content.
Capability inventory: The skill utilizes a local Python script that performs Git operations via subprocess calls.
Sanitization: The translation-drift-check.py script implements input validation by verifying that source_commit values match an expected hash format before use in command arguments.

docs-localization-strategy