convex-backup

Pass

Audited by Gen Agent Trust Hub on Aug 12, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes data from external Convex database environments to verify restore success, creating an indirect prompt injection surface.
  • Ingestion points: Data is read from the database using npx convex export, the tables MCP tool, and runOneoffQuery.
  • Boundary markers: There are no explicit delimiters or specific instructions for the agent to disregard potential instructions embedded within the database rows being verified.
  • Capability inventory: The agent has the ability to execute shell commands (npx convex) and interact with database APIs.
  • Sanitization: The instructions do not mention sanitizing or filtering the database content before the agent analyzes it for verification purposes.
  • [COMMAND_EXECUTION]: The skill uses npx convex to perform data exports, deployment of preview environments, and data imports. These commands are integral to the backup and recovery workflow provided by the vendor.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 12, 2026, 03:15 PM
Security Audit — agent-trust-hub — convex-backup