cloudbase-agent-ts

Pass

Audited by Gen Agent Trust Hub on Feb 22, 2026

Risk Level: SAFEPROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
  • PROMPT_INJECTION (LOW): Indirect Prompt Injection Surface identified. * Ingestion points: User messages are ingested via the RunAgentInput interface and HTTP headers (Authorization JWT) as seen in adapter-langchain.md and adapter-langgraph.md. * Boundary markers: Implementation examples do not demonstrate explicit delimiters or system instructions to ignore embedded commands. * Capability inventory: The system supports server-side tool execution and database persistence via CloudBaseSaver. * Sanitization: No explicit sanitization or validation of incoming message content is demonstrated in the provided code snippets.
  • EXTERNAL_DOWNLOADS (LOW): The skill requires installation of multiple packages from the @cloudbase and @ag-ui scopes. While these organizations are not in the predefined 'Trusted' list, they are legitimate components of the Tencent CloudBase ecosystem.
  • DATA_EXFILTRATION (SAFE): No patterns of unauthorized data access or exfiltration were detected. JWT decoding via atob is used for standard multi-tenant isolation and authentication purposes.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 22, 2026, 07:54 PM