The Agent Skills Directory

[SAFE]: The skill is designed to perform precise translations from multiple source languages into Chinese, prioritizing structural fidelity over linguistic fluency. The instructions are comprehensive and focused entirely on the translation task.
[COMMAND_EXECUTION]: The workflow utilizes local utilities such as grep and Python scripts for technical audits (e.g., counting punctuation, checking term consistency). These tools are described as project-specific components resident within the user's workspace, rather than remotely fetched or dynamically generated malicious code.
[DATA_EXFILTRATION]: The agent is instructed to manage a local workspace structure (source/, segments/, translated/) to facilitate translation and session recovery. The analysis found no evidence of network operations, access to sensitive system paths (like .ssh or .aws), or hardcoded credentials.
[PROMPT_INJECTION]: While the skill ingests external source text, which technically constitutes an indirect prompt injection surface, it provides a rigorous 'Constitution' with three specific 'Laws' that mandate verbatim fidelity. These rules serve to isolate the source text as data to be transformed rather than instructions to be followed, minimizing the risk of the agent adopting malicious commands embedded in the source material.

apostle-translation