blocking-muting-management

Pass

Audited by Gen Agent Trust Hub on Feb 28, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTIONDATA_EXFILTRATIONNO_CODE
Full Analysis
  • [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection because it processes untrusted data from external social media profiles to trigger automated account actions.\n
  • Ingestion points: The scripts read content from x.com/USERNAME/followers, user timelines, and search results as described in SKILL.md.\n
  • Boundary markers: No boundary markers or instructions to ignore embedded commands are mentioned in the detection logic.\n
  • Capability inventory: The documented scripts can perform automated block, unblock, mute, unmute, and report spam actions using DOM selectors like [data-testid="block"].\n
  • Sanitization: The bot detection heuristics (e.g., bio keywords) lack sanitization or validation of the input content.\n- [COMMAND_EXECUTION]: The documentation describes scripts that programmatically interact with the browser's Document Object Model (DOM) to automate navigation and click-based actions on the X platform.\n- [DATA_EXFILTRATION]: The scripts read and process user-specific data, including follower lists, account metadata (age, follower counts), and tweet content to execute its management functions.\n- [NO_CODE]: The skill references multiple JavaScript files in a src/ directory (e.g., massBlock.js, blockBots.js, muteByKeywords.js) and a global window.XActions object, but the source code for these scripts is not included in the provided files.
Audit Metadata
Risk Level
SAFE
Analyzed
Feb 28, 2026, 09:53 AM