The Agent Skills Directory

[Data Exposure & Exfiltration] (SAFE): The i18n_checker.py script reads local project files for analysis but does not transmit data externally or access sensitive system directories like ~/.ssh or ~/.aws.
[Indirect Prompt Injection] (SAFE): The script ingests untrusted data (user code and JSON locale files) to perform regex-based auditing. It lacks dangerous sinks such as shell execution, dynamic code evaluation (eval/exec), or network requests, meaning it cannot be used as a vector for malicious payloads.
[Remote Code Execution] (SAFE): No patterns for downloading or executing remote code (e.g., curl | bash) were found. The skill relies on standard Python libraries (pathlib, re, json).
[Command Execution] (SAFE): The script performs filesystem reads using pathlib but does not invoke system shells or subprocesses.
[Metadata Poisoning] (SAFE): The skill metadata correctly reflects the tool's purpose and does not contain hidden instructions or deceptive descriptions.

i18n-localization