The Agent Skills Directory

[COMMAND_EXECUTION]: The skill uses shell commands via Python and standard utilities (e.g., python3 -c "from pathlib import Path; print(len(Path('path').read_text()))" and wc -m) to accurately count characters and verify file integrity during the writing process.
[EXTERNAL_DOWNLOADS]: It leverages a specialized story-researcher agent to gather external information on historical or geographic facts, which involves retrieving data from the network and saving it to the local 参考资料/ directory.
[PROMPT_INJECTION]: Reference documents such as anti-ai-writing.md provide detailed instructions on 'de-AI' techniques to strip common LLM linguistic fingerprints and bypass AI-detection styles. While used here for literary quality, these represent instructions aimed at modifying core model output patterns.
[SAFE]: All identified command execution and network operations are intrinsic to the novel writing and research workflow. The skill adheres to a strict project directory structure, and no evidence of data exfiltration, credential theft, or malicious obfuscation was found across the 38 analyzed files.

story-long-write