The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill is designed to process user-supplied screenplays and scripts to generate production assets and model prompts.
Ingestion points: Multiple files including SKILL.md, references/asset-first-pipeline.md, and references/screenplay-gate-engine.md define the workflow for ingesting external script data.
Boundary markers: Absent. While the skill defines rigid structural sections, it lacks explicit instructions for the AI to ignore or neutralize potentially malicious instructions embedded within the processed screenplay data.
Capability inventory: The skill possesses capabilities for file system operations (via the agent) and local command execution through included scripts.
Sanitization: No explicit sanitization or filtering logic for untrusted text data is provided in the instruction set.
[DYNAMIC_EXECUTION]: The skill includes Python scripts that perform local execution or generate dynamic content.
Subprocess Execution: The script scripts/check_package.py uses subprocess.run() to execute another local script, scripts/validate_prompt.py, as part of its integrity check routine.
Script Generation: The script scripts/build_board_lite.py dynamically compiles markdown data into a single-file HTML storyboard containing embedded JavaScript and CSS.

short-drama-director