bilibili-keyword-accounts

Pass

Audited by Gen Agent Trust Hub on Jul 1, 2026

Risk Level: SAFECOMMAND_EXECUTIONDATA_EXFILTRATIONPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes a local Python script search_bilibili_user.py to facilitate search requests.
  • [DATA_EXFILTRATION]: The script performs network requests to doubaoya.com. It transmits the authentication token stored in the DOUBAOYA_API_KEY environment variable to this domain. This represents communication with the vendor's intended service infrastructure.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes and displays user-generated content from external Bilibili profiles.
  • Ingestion points: Untrusted data enters the context through account names and descriptions in the API response within search_bilibili_user.py.
  • Boundary markers: The skill instructions provide a Markdown table structure for output, though no explicit 'ignore instructions' guards are present.
  • Capability inventory: The skill allows script execution and data rendering.
  • Sanitization: Content from the sign and uname fields is displayed without prior filtering or validation.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 1, 2026, 05:47 PM
Security Audit — agent-trust-hub — bilibili-keyword-accounts