The Agent Skills Directory

[COMMAND_EXECUTION]: The script scripts/batch_transcribe.py uses subprocess.run to call transcribe.py for each entry in the input file. Although it uses argument lists to prevent shell injection, it represents dynamic process creation based on external file content.- [REMOTE_CODE_EXECUTION]: The scripts/transcribe.py script loads a transcription model from ModelScope with trust_remote_code=True. This allows the execution of arbitrary Python code included in the remote model repository (iic/SenseVoiceSmall) during the model initialization process.- [EXTERNAL_DOWNLOADS]: The skill downloads video data, audio, and subtitles from Bilibili, and fetches machine learning models from ModelScope. These are functional requirements but involve external network dependencies.- [INDIRECT_PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection by processing untrusted video metadata and subtitles.
Ingestion points: scripts/transcribe.py parses Bilibili video titles and subtitle files.
Boundary markers: No delimiters or safety instructions are used when interpolating external content into the Markdown output.
Capability inventory: The skill can execute subprocesses and perform network requests.
Sanitization: While filenames are sanitized, the primary content of the transcribed Markdown is not filtered for malicious instructions.- [DYNAMIC_EXECUTION]: The initialization of the funasr model involves loading and executing code from a remote repository, which is a form of dynamic execution.

bilibili-transcribe