The Agent Skills Directory

[SAFE]: The skill utilizes local Python utility scripts (scripts/episode_intake.py) to manage and index multi-episode screenplay sources. Analysis of these scripts confirms they perform standard file I/O using the Python standard library (argparse, json, os, pathlib) to calculate byte spans and slice text. No network communication or unauthorized system modifications are present.\n- [SAFE]: The skill handles user-provided drama materials (novels, scripts) from the 输入/ directory. It implements a 'mechanical intake' pattern (documented in references/multi-episode-intake.md) where the episode_intake.py script records byte spans to create an index (episode-intake-index.json). This ensures that only specific slices of content are processed at a time, providing a clear boundary against indirect prompt injection and managing large context inputs safely.\n- [SAFE]: No obfuscated URLs, hardcoded credentials, or unauthorized remote code execution patterns were detected. All project management commands described in the documentation are intended for local execution by the user for managing the drama development lifecycle.

short-drama-develop