The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides specific command-line instructions in the SKILL.md documentation to run local Python scripts (screenplay_index.py, duration_estimate.py, selftest.py). These scripts are used for technical tasks such as indexing scene metadata and calculating script duration based on character counts and pacing rates.
[DYNAMIC_EXECUTION]: The skill involves the dynamic execution of provided Python utility scripts. The agent is instructed to use a Python one-liner to generate a unique temporary file path and then pass that path as an argument to analysis scripts that process local Markdown content.
[INDIRECT_PROMPT_INJECTION]: The skill has a defined attack surface for indirect prompt injection as it is designed to ingest and process untrusted user-supplied content, including drama outlines and Markdown-formatted scripts.
Ingestion points: Ingests user-provided text files, including 剧本.md and related outlines, for revision and analysis.
Boundary markers: The skill enforces a specific structural grammar for screenplays (e.g., # EP[ID] for episodes and ## [ID] [Space] · [Location] · [Time] for scenes) which acts as a structural boundary.
Capability inventory: The agent uses shell commands to execute Python scripts that perform file reading, pattern matching via regular expressions, and metadata writing to the system's temporary directory.
Sanitization: Input data is parsed using regular expressions and standard JSON libraries to extract specific screenplay blocks, reducing the likelihood of the model following instructions embedded within the processed text.

short-drama-write