What it does
Scans page source and network requests to list every externally hosted script (analytics, chat widgets, ad networks, fonts, embeds), attributes each one's load time and render-blocking behavior, and flags scripts that could reasonably be deferred, lazy-loaded, or removed outright.
Install
This skill isn't packaged in an installable registry. It's a plain SKILL.md file you can copy directly. Copy it from the SKILL.md tab and save it at the path for your assistant:
Use it in your AI assistant
Claude Code
.claude/skills/third-party-script-auditor/ (or ~/.claude/skills/third-party-script-auditor/ for all projects)
GitHub Copilot (VS Code)
.github/skills/third-party-script-auditor/
OpenAI Codex
.codex/skills/third-party-script-auditor/
Cursor
.cursor/skills/third-party-script-auditor/ (or ~/.cursor/skills/third-party-script-auditor/ globally)
Compatibility
Any WordPress site; a browser network trace or a tool like WebPageTest gives the most accurate script inventory.
The complete SKILL.md for this skill:
---
name: third-party-script-auditor
description: Use when a user wants to know what third-party scripts are loaded on their site, or suspects a widget, tracker, or embed is slowing down page load.
---
# Third-Party Script Auditor
## When to use
The site feels slower than it should and the user suspects third-party scripts (analytics, chat widgets, ad tags, embeds) are the cause, or they just want a full inventory of what's being loaded and from where.
## Procedure
1. List every externally hosted script loaded on representative pages: analytics/tracking, chat widgets, ad network tags, font providers, social embeds, A/B testing tools.
2. For each script, note its load timing (blocking vs. async/deferred) and its measured impact on page load if that data is available from a network trace or performance report.
3. Flag scripts that block rendering unnecessarily and could be safely switched to async or deferred loading without breaking functionality.
4. Flag scripts that appear to duplicate functionality (e.g. two analytics tools tracking the same thing) as removal candidates.
5. Flag scripts that seem abandoned, no longer tied to an active feature or campaign, for the user to confirm whether they're still needed.
6. Present the inventory as a table: script, purpose, loading behavior, and a recommendation (keep as is, defer, lazy-load, or remove pending confirmation).
## Guardrails
- Never remove or modify script loading behavior directly, only report the inventory and recommendations for the user to implement and test.
- Do not recommend removing a script tied to a business function (payment processing, required legal tracking, consent management) without flagging that its removal needs sign-off first.