skills/simota/agent-skills/dock/Gen Agent Trust Hub

dock

Pass

Audited by Gen Agent Trust Hub on Jul 27, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill exclusively uses official Apple developer tools (xcodebuild, codesign, notarytool, and spctl) and well-known, industry-standard libraries like the Sparkle framework for updates. These are legitimate utilities for software development and distribution.
  • [SAFE]: The documentation and core contract emphasize security-first principles, specifically mandating 'Sandbox by default, least privilege' and proper notarization for all builds. The 'sandbox' recipe provides detailed guidance on least-privilege entitlement selection.
  • [SAFE]: Persistence mechanisms are handled through modern, system-sanctioned APIs like SMAppService, which replaces legacy, less secure methods (like SMJobBless) and ensures user visibility and control in System Settings.
  • [PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface as it ingests data from various collaboration partners.
  • Ingestion points: Data enters the agent context through several handoff files including FORGE_TO_DOCK_HANDOFF (prototypes), VISION_TO_DOCK_HANDOFF (design direction), and BUILDER_TO_DOCK_HANDOFF (API specifications).
  • Boundary markers: The instructions do not define specific delimiters or 'ignore' instructions for these ingestion points, although they are processed within a structured workflow.
  • Capability inventory: The skill has broad capabilities including file system access, network operations, and code generation, all of which are necessary for its role as a professional development tool.
  • Sanitization: While specific input sanitization is not detailed, the skill includes a mandatory 'VERIFY' phase that audits the generated code for entitlement compliance, sandbox safety, and platform constraints.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 27, 2026, 01:12 PM
Security Audit — agent-trust-hub — dock