The Agent Skills Directory

[NO_CODE]: The skill references several critical JavaScript files including src/notificationManager.js, scripts/scrapeNotifications.js, src/welcomeNewFollowers.js, and src/engagementLeaderboard.js. None of these files are included in the provided package, meaning the actual logic being executed cannot be audited for malicious behavior.
[INDIRECT_PROMPT_INJECTION]: The skill is designed to scrape and process external, untrusted content from X/Twitter notifications.
Ingestion points: Data is collected from DOM elements matching [data-testid="notification"] and [data-testid="notification"] [dir="auto"] as specified in the DOM Selectors table.
Boundary markers: There are no boundary markers or instructions to ignore embedded prompts mentioned in the documentation.
Capability inventory: The skill performs bulk management, notification scraping, and automated follower engagement (sending welcome messages).
Sanitization: There is no evidence of sanitization or validation logic to handle potentially malicious notification content.
[DYNAMIC_EXECUTION]: The documentation explicitly instructs users to open browser DevTools and paste scripts into the console. This 'Self-XSS' pattern is a significant security risk, as users are guided to bypass platform security controls and execute arbitrary JavaScript in the context of their active session.
[METADATA_POISONING]: The skill metadata identifies the author as 'nichxbt', which does not match the expected author context 'nirholas' provided for this analysis, representing a potential discrepancy in identity or ownership.

notifications-management