cross-domain-borrow

Pass

Audited by Gen Agent Trust Hub on Aug 23, 2026

Risk Level: SAFE
Full Analysis
  • [PROMPT_INJECTION]: The skill contains strict formatting instructions (e.g., '禁止改写', '严格照提示语') intended to ensure the agent follows the specific problem-solving framework and provides the raw template when requested. These are standard prompt engineering techniques to maintain consistency and do not attempt to bypass safety filters or override system-level security constraints.
  • [COMMAND_EXECUTION]: The documentation mentions a synchronization script (scripts/sync.py) and a verification process for internal development workflows. However, the skill itself does not contain any code that executes this script or any other shell commands at runtime.
  • [INDIRECT_PROMPT_INJECTION]: The skill defines a clear ingestion point for untrusted user data within the template placeholder 【说明背景、当前做法、现实约束和具体卡点】.
  • Ingestion points: User input provided to the prompt template in SKILL.md.
  • Boundary markers: The template uses square brackets 【 】 as delimiters.
  • Capability inventory: The skill is limited to text generation. No subprocess calls, file writes, or network operations are present in the provided files.
  • Sanitization: No explicit sanitization or filtering of user input is implemented, but the lack of dangerous capabilities makes this a safe interaction surface.
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 23, 2026, 06:05 PM
Security Audit — agent-trust-hub — cross-domain-borrow