Convex Agents Streaming
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [Prompt Injection] (LOW): The skill provides patterns for handling user prompts that create an attack surface for indirect prompt injection.\n
- Ingestion points: The
streamResponseaction inSKILL.mdaccepts apromptargument from the user.\n - Boundary markers: No delimiters or safety instructions are included in the code examples to isolate user input from agent instructions.\n
- Capability inventory: The skill uses
@convex-dev/agentfor streaming; it does not involve subprocess execution, file-writing, or network requests to external domains.\n - Sanitization: There is no evidence of input sanitization or validation in the provided code snippets.\n- [Data Exposure & Exfiltration] (SAFE): No credentials or sensitive data access patterns identified.\n- [Remote Code Execution] (SAFE): No remote code execution or dynamic code evaluation detected.
Audit Metadata