convex-components
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONPROMPT_INJECTION
Full Analysis
- EXTERNAL_DOWNLOADS (LOW): The skill references several npm packages within the
@convex-devscope. While this scope is not on the explicit whitelist, these packages are standard for Convex development, and the skill purpose is centered around this ecosystem. - COMMAND_EXECUTION (LOW): The setup process requires executing
npx convex dev. This is a routine part of the Convex development workflow for generating code and synchronizing the backend schema. - PROMPT_INJECTION (LOW): There is a potential for indirect prompt injection in the Agent and RAG components where LLMs process user inputs. Findings: 1. Ingestion points:
agent.mdandrag.mdaccept prompt arguments. 2. Boundary markers: Not specifically implemented in the provided code snippets. 3. Capability inventory: The skill utilizesctx.runMutationwhich allows database writes. 4. Sanitization: The instructions explicitly require validation for all public function arguments.
Audit Metadata