aidlc-prototype

Warn

Audited by Gen Agent Trust Hub on Jul 20, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: The skill executes shell commands to install third-party libraries and scaffold project environments based on specifications found in external files.
  • [EXTERNAL_DOWNLOADS]: During the build process, the skill downloads and installs various software packages (e.g., Express, Fastify, Expo) from public registries.
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from requirements and design files to determine its execution logic.
  • Ingestion points: Reads requirements.md, blueprints/resources.md, and design documentation to scope the prototype.
  • Boundary markers: Absent; the instructions do not include delimiters or warnings to ignore embedded commands within the requirement stories.
  • Capability inventory: Includes shell command execution for package management and extensive file system read/write operations.
  • Sanitization: No validation or sanitization of library names or story content is performed before they are used to generate commands.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Jul 20, 2026, 08:00 AM
Security Audit — agent-trust-hub — aidlc-prototype