Skills
Navigate
skills/different-ai/agent-bank/tweet-rl-tracker/Gen Agent Trust Hub

tweet-rl-tracker

Pass

Audited by Gen Agent Trust Hub on Sep 20, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONDATA_EXFILTRATION
Full Analysis
  • [INDIRECT_PROMPT_INJECTION]: The skill processes data from external Twitter URLs which could contain malicious instructions designed to influence the agent. Evidence: The extraction of tweetText and author metadata via browser scripts. Ingestion points: Data extracted from x.com. Capability inventory: Local file writing and Notion database interactions. Boundary markers: None. Sanitization: None.
  • [DYNAMIC_EXECUTION]: The skill provides specific code snippets for the agent to execute during runtime for browser interaction and file handling. Evidence: Use of evaluate_script for DOM manipulation and Node.js require('fs') for managing temporary files in /tmp.
  • [DATA_EXFILTRATION]: The instructions suggest the agent upload captured data to third-party services like S3 or Cloudinary to facilitate Notion integration. Evidence: Guidance in the 'Saving Screenshot to Notion' section directing the agent to use external hosting services.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 20, 2026, 06:00 AM