saas-scaffolder

Pass

Audited by Gen Agent Trust Hub on Sep 8, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes external data from a "product brief" and existing repository files (e.g., package.json, go.mod) to guide its scaffolding logic.
  • Ingestion points: Validated product brief and local stack discovery via rg commands in SKILL.md.
  • Boundary markers: No specific boundary markers or "ignore" instructions are defined for the input brief data to prevent accidental instruction execution.
  • Capability inventory: The skill utilizes shell commands (rg, <package-manager> lint/test/build) and has the capability to write new source files to the disk.
  • Sanitization: There is no explicit sanitization or validation defined for the content of the product brief before it influences the agent's actions.
  • [COMMAND_EXECUTION]: The skill employs system commands for repository inspection and project verification.
  • Evidence: Use of rg (ripgrep) to search for configuration files and keywords like "auth" or "stripe", and execution of package manager commands (lint, typecheck, test, build) to verify the generated code slice.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 8, 2026, 06:56 AM
Security Audit — agent-trust-hub — saas-scaffolder