baoyu-post-to-x

Pass

Audited by Gen Agent Trust Hub on Nov 2, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTIONDYNAMIC_EXECUTION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied Markdown files to generate posts and articles, which represents an indirect injection surface if the content contains instructions intended for the AI agent.
  • Ingestion points: Markdown files processed by scripts/md-to-html.ts and remote images potentially fetched via vendor libraries.
  • Boundary markers: The content is parsed and converted to HTML, but no explicit 'ignore embedded instructions' prompts are visible in the processing chain.
  • Capability inventory: The skill can execute shell commands (bun, osascript, powershell), interact with the system clipboard, and control a browser session with active user credentials.
  • Sanitization: Uses standard Markdown parsing libraries (marked, unified) and includes basic HTML escaping.
  • [COMMAND_EXECUTION]: The skill executes various system commands and automation tools to perform browser control and clipboard operations.
  • Evidence: Uses osascript (macOS), powershell.exe (Windows), and xdotool/ydotool (Linux) in scripts/paste-from-clipboard.ts and scripts/check-paste-permissions.ts to simulate user input.
  • Evidence: Uses spawn and spawnSync to execute swift, npx, and bun scripts.
  • [DYNAMIC_EXECUTION]: The skill generates and executes temporary automation scripts at runtime to provide cross-platform support.
  • Evidence: scripts/copy-to-clipboard.ts and scripts/check-paste-permissions.ts dynamically generate Swift source code, write it to temporary files, and execute it using the swift interpreter to access macOS AppKit clipboard APIs.
  • Evidence: Dynamically constructs AppleScript and PowerShell command strings to interact with the operating system's UI and clipboard.
Audit Metadata
Risk Level
SAFE
Analyzed
Nov 2, 2026, 09:45 AM