feature-design-assistant

Pass

Audited by Gen Agent Trust Hub on Apr 30, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill provides a collaborative framework for designing software features. It primarily uses user interaction for requirement gathering and writes documentation to the 'docs/designs/' directory, which is standard behavior for a project assistant.
  • [DATA_EXFILTRATION]: The skill discovery phase involves reading codebase files, and Phase 5 involves writing design specifications to the local file system. These operations are performed within the local project scope and do not involve unauthorized data transmission to external domains or access to sensitive credentials.
  • [PROMPT_INJECTION]: The skill ingests codebase content during context discovery, which is a potential surface for indirect prompt injection. However, the skill's capabilities are limited to generating documentation and task lists that are subject to user review, minimizing the risk of unauthorized command execution.
  • Ingestion points: Phase 1: Context Discovery (reads codebase files).
  • Boundary markers: Absent; the skill does not use specific delimiters to isolate external code content from instructions.
  • Capability inventory: File writing to documentation paths ('docs/designs/*.md').
  • Sanitization: Relies on the user for validation and review before proceeding to implementation.
Audit Metadata
Risk Level
SAFE
Analyzed
Apr 30, 2026, 10:00 PM
Security Audit — agent-trust-hub — feature-design-assistant