domain-scaffolder-backend

Pass

Audited by Gen Agent Trust Hub on Mar 10, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill reads and processes plan files (shared.md, backend.md, schema.mmd) in Step 1 to derive business logic and code structure, creating an indirect prompt injection surface. * Ingestion points: Plan files at {plan_root}/{slice}/ are read in Step 1. * Boundary markers: None specified to delimit plan data from instructions. * Capability inventory: The skill can write multiple source files (Steps 3-9) and execute shell commands (Steps 10-11). * Sanitization: No validation or sanitization of input plan content is performed.
  • [COMMAND_EXECUTION]: The skill executes shell commands for migrations and tests in Steps 10 and 11 using tool paths and commands defined in mode configuration files. These commands interpolate the {slice} variable from the plan, which could lead to command injection if the plan name or content is manipulated to include shell metacharacters.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 10, 2026, 02:39 PM