baoyu-danger-x-to-markdown

Warn

Audited by Gen Agent Trust Hub on Feb 24, 2026

Risk Level: MEDIUMCOMMAND_EXECUTIONCREDENTIALS_UNSAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
  • [COMMAND_EXECUTION]: In scripts/cookies.ts, the skill identifies and executes a local browser binary (Chrome, Chromium, or Edge) with a remote debugging port enabled. This is used to programmatically extract authentication cookies via the Chrome DevTools Protocol (CDP).
  • [CREDENTIALS_UNSAFE]: The skill manages sensitive X authentication data. It reads tokens from environment variables (X_AUTH_TOKEN, X_CT0) and persists them to a local JSON file (cookies.json) in the user's application support directory.
  • [EXTERNAL_DOWNLOADS]: The skill fetches media assets (images and videos) from X.com's well-known media servers (pbs.twimg.com, video.twimg.com) to save them locally.
  • [PROMPT_INJECTION]: The skill exhibits an indirect prompt injection surface as it retrieves and processes arbitrary content from X.com.
  • Ingestion points: Data enters the context via scripts/graphql.ts, which fetches tweet and article text from external API responses.
  • Boundary markers: Absent. The resulting markdown does not use delimiters to clearly isolate untrusted content from the agent's instructions.
  • Capability inventory: Subprocess execution (browser spawning), file system read/write access, and network operations.
  • Sanitization: Absent. Text retrieved from X.com is formatted into markdown without being sanitized for potential instructions that could influence agent behavior.
Audit Metadata
Risk Level
MEDIUM
Analyzed
Feb 24, 2026, 03:29 AM