The Agent Skills Directory

[DYNAMIC_EXECUTION]: The script scripts/timeline_lib.py utilizes the eval() function to process and interpret expressions extracted from TypeScript configuration files (timeline.ts). This allows the skill to dynamically load and compute video timing and metadata, but it introduces a risk of code execution if the source files are compromised.
[INDIRECT_PROMPT_INJECTION]: The skill's workflow includes fetching historical records and text from external websites like zh.wikisource.org to generate video scripts. This creates a surface for indirect prompt injection where instructions hidden in the historical data could influence the agent's behavior.
Ingestion points: External historical text is fetched using curl and integrated into the script generation process in references/选题与史料.md.
Boundary markers: No specific delimiters or safety warnings are implemented to isolate external content from the agent's instructions.
Capability inventory: The skill includes capabilities for command execution via subprocess, file writing, and network operations.
Sanitization: The scripts do not perform explicit sanitization or filtering of the external data before processing.
[COMMAND_EXECUTION]: The Python script scripts/beat_accents.py uses subprocess.run() to execute system-level commands, specifically invoking ffmpeg to analyze audio frequency and intensity for video synchronization.

weilanx-creator-lamplight-video