Convex Agents Tools
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [Indirect Prompt Injection] (LOW): The skill enables autonomous tool calling based on untrusted user input, creating a surface for indirect prompt injection.
- Ingestion points: The
requestparameter in theautonomousAgentaction inSKILL.mdis passed directly to the LLM. - Boundary markers: No specific delimiters or 'ignore embedded instructions' warnings are present in the provided code snippet.
- Capability inventory: The example tool
getUserDataToolallows database read access (ctx.runQuery). - Sanitization: The skill demonstrates the use of Zod (
z.string().email()) for strict schema validation and description-based constraints on tool arguments, which mitigates basic injection attempts into the tool parameters.
Audit Metadata