baoyu-post-to-wechat

Fail

Audited by Gen Agent Trust Hub on Feb 16, 2026

Risk Level: HIGHPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [Indirect Prompt Injection] (HIGH): The skill creates a significant attack surface by reading external markdown files and command-line arguments to perform automated actions on a live social media account. 1. Ingestion points: markdown files (article.md) and CLI parameters (--title, --content). 2. Boundary markers: Absent; no delimiters are used to separate user data from instructions. 3. Capability inventory: Uses Chrome CDP for full browser automation and session-based posting. 4. Sanitization: None described; content is directly converted and posted.
  • [Command Execution] (MEDIUM): The skill executes scripts using 'npx -y bun', which involves runtime execution of TypeScript files located in the skill's subdirectory.
Recommendations
  • AI detected serious security threats
Audit Metadata
Risk Level
HIGH
Analyzed
Feb 16, 2026, 10:57 AM