The Agent Skills Directory

Prompt Injection (SAFE): No patterns found attempting to override agent instructions or bypass safety filters.
Data Exposure & Exfiltration (SAFE): No sensitive file access, hardcoded credentials, or network requests detected. The script is designed to read only specific code and localization file extensions.
Remote Code Execution (SAFE): No use of eval(), exec(), or subprocess. The script relies entirely on Python's standard library for file traversal and regular expression matching.
Indirect Prompt Injection (LOW): The i18n_checker.py script reads external file content from a user-specified directory. Evidence Chain: 1. Ingestion points: scripts/i18n_checker.py reads local code and JSON files. 2. Boundary markers: Absent. 3. Capability inventory: The script only prints results to the console; it has no file-write or network capabilities. 4. Sanitization: Output is limited to first 40 characters of regex matches to prevent large-scale data ingestion. Severity is LOW as it is a diagnostic tool.
Persistence & Privilege Escalation (SAFE): No attempts to modify system configurations, shell profiles, or acquire elevated permissions.

i18n-localization