yichen-bookmarks-export

Pass

Audited by Gen Agent Trust Hub on Aug 19, 2026

Risk Level: SAFEPROMPT_INJECTION
Full Analysis
  • [PROMPT_INJECTION]: The skill operates on bookmarks retrieved from external social media platforms, creating an indirect prompt injection attack surface.
  • Ingestion points: The skill ingests bookmark data and URLs from Xiaohongshu, Douyin, and X via an underlying exporter tool (SKILL.md).
  • Boundary markers: It employs a strict "this-turn authorization" gate (当轮授权闸门) which requires fresh user consent for every session and explicitly instructs the agent to ignore requests for unauthorized platforms or activities like media downloading (SKILL.md).
  • Capability inventory: The skill is capable of writing link files to local directories and invoking the yichen-social-bookmarks-exporter skill (SKILL.md).
  • Sanitization: The skill architecture relies on platform-specific routing logic and mentions URL validation within its underlying collectors to filter invalid or malicious links (SKILL.md, tests/routes.json).
Audit Metadata
Risk Level
SAFE
Analyzed
Aug 19, 2026, 12:57 AM
Security Audit — agent-trust-hub — yichen-bookmarks-export