doubao-chat

Pass

Audited by Gen Agent Trust Hub on Sep 19, 2026

Risk Level: SAFEDATA_EXFILTRATIONINDIRECT_PROMPT_INJECTION
Full Analysis
  • [DATA_EXFILTRATION]: The script scripts/chat.js interacts with an external API hosted on Vercel (doubao-free-api.vercel.app), a well-known service. It transmits the user-provided DOUBAO_SESSIONID in the Authorization header to authenticate these requests.\n- [INDIRECT_PROMPT_INJECTION]: The script scripts/chat.js processes user input from command-line arguments and interpolates it directly into a chat completion request without sanitization or boundary markers. This could allow malicious input to influence the AI's behavior.\n
  • Ingestion points: User input is read from process.argv[2] in scripts/chat.js.\n
  • Boundary markers: The script does not use delimiters or specific instructions to separate user input from the prompt context.\n
  • Capability inventory: The script performs network requests to an external API.\n
  • Sanitization: No validation or filtering is applied to the input string before processing.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 19, 2026, 11:32 PM
Security Audit — agent-trust-hub — doubao-chat