dbs-video-extract

Pass

Audited by Gen Agent Trust Hub on Sep 2, 2026

Risk Level: SAFECOMMAND_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill uses subprocess.run within scripts/tikhub_api.py and scripts/extract_video_transcript.py to call the macOS security command. This is used exclusively for retrieving API keys from the system keychain in a secure manner.
  • [DATA_EXPOSURE]: The skill reads and writes service credentials to a local configuration file located at ~/.config/dbs/API_Keys.md. It implements best practices by restricting file permissions to 0o600 (owner-only access) and providing a script (scripts/configure_api_key.py) that uses getpass to read credentials without echoing them to the terminal.
  • [EXTERNAL_DOWNLOADS]: The skill communicates with tikhub.io and qingdou.vip APIs to perform its primary function of fetching video metadata and transcripts. These are external service dependencies required for the skill to operate.
  • [INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data in the form of short video URLs and sharing text from external platforms.
  • Ingestion points: User-supplied links and text are ingested via command-line arguments or standard input in scripts/extract_video.py.
  • Boundary markers: None explicitly defined in the scripts, as the input is treated as a single data parameter for API submission.
  • Capability inventory: The skill possesses network communication capabilities and file write permissions to save results.
  • Sanitization: The scripts/extract_video_transcript.py script includes a safe_name function that uses regular expressions to strip illegal characters (such as path separators) and enforces length limits, effectively preventing path traversal or directory manipulation attacks.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 2, 2026, 09:23 AM
Security Audit — agent-trust-hub — dbs-video-extract