Convex Agents Context
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- Indirect Prompt Injection (LOW): The skill provides mechanisms to inject message history, user memories, and search results into the LLM context. 1. Ingestion points: External data enters through searchOtherThreads and contextHandler in SKILL.md. 2. Boundary markers: Absent. The examples do not demonstrate delimiters or instructions to ignore embedded commands in the injected context. 3. Capability inventory: The skill triggers LLM text generation using the myAgent.generateText method. 4. Sanitization: Absent. Injected data is interpolated directly into the context array without escaping or validation.
Audit Metadata