bilibili-subtitle-downloader

Pass

Audited by Gen Agent Trust Hub on Mar 6, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes external subtitle data as input for AI summarization, creating a surface for indirect prompt injection.
  • Ingestion points: Subtitle content is retrieved from remote Bilibili URLs in scripts/download_and_chunk.py and scripts/cheese_downloader.py.
  • Boundary markers: Sub-agent instructions in SKILL.md do not utilize delimiters or warnings to ignore embedded commands.
  • Capability inventory: The agent reads locally stored subtitle chunks and summarizes them using the provided prompts.
  • Sanitization: No content filtering or validation is performed on the subtitle data prior to processing by the sub-agent.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 6, 2026, 09:56 AM