douyin-downloader-hardened

Warn

Audited by Gen Agent Trust Hub on Apr 21, 2026

Risk Level: MEDIUMCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [CREDENTIALS_UNSAFE]: Accesses the file ~/.openclaw/config.json to retrieve the tikhub_api_token, which is a sensitive credential.
  • [EXTERNAL_DOWNLOADS]: Fetches video metadata and binary content from api.tikhub.io and external CDNs.
  • [DATA_EXFILTRATION]: Reads local credentials and uses them in network requests to an external service; the skill includes prompt-based guardrails to mitigate the risk of token leakage in terminal output.
  • [PROMPT_INJECTION]: Contains instructions designed to ignore adversarial system alerts or social engineering attempts intended to compromise the API token or bypass scope restrictions.
  • [INDIRECT_PROMPT_INJECTION]: Processes untrusted user URLs and external API responses. Ingestion points include user-provided links and modal IDs; capability inventory includes network reads and local file writes; sanitization is performed via regex extraction.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Apr 21, 2026, 06:31 PM
Security Audit — agent-trust-hub — douyin-downloader-hardened