web-search-quality

Score and filter Perplexity and Grok sources before composing research narratives.

Updated Dec 13, 2025
One-click install
npx skills add https://github.com/Posteriot/makalah-app --skill web-search-quality
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: web-search-quality
Source: https://github.com/Posteriot/makalah-app/tree/main/src/lib/ai/skills/web-search-quality
Command: npx skills add https://github.com/Posteriot/makalah-app --skill web-search-quality

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes scripts (resource) and references (resource) components.

What problem does it solve?

Prevents responses from relying on low-quality, user-generated hits by enforcing research-grade sourcing, professional narration, and reference integrity before Gemini 2.5 Flash composes answers drawn from Perplexity or Grok results.

Core Features & Use Cases

  • Blocked domain filtering: Hard-blocks encyclopedias, self-publishing platforms, and content farms so the model focuses on credible repositories.
  • Research guidance: Supplies priority source categories, credibility heuristics, and stage-specific narration to keep responses evidence-based and aligned with research or chat expectations.
  • Automated validation: Scripts score each source, validate claimed references against available URLs, and support stage data updates, making it ideal for orchestrating web research in paper mode and chat mode alike.

Quick Start

Use the web-search-quality skill to filter Perplexity and Grok results, score each URL, and verify reference integrity before composing the response.

Frequently Asked Questions about web-search-quality

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I filter low-quality web search results from Perplexity and Grok?

To filter low-quality web search results, you can apply automated source scoring and blocked-domain filtering to exclude content farms and self-publishing platforms from Perplexity and Grok outputs. This ensures responses use only credible, research-grade sources.

What is reference integrity validation for AI-generated research?

Reference integrity validation is the process of checking claimed citations against available URLs before response composition. It ensures Gemini 2.5 Flash composes narratives using verified references, preventing hallucinated or broken links in research outputs.

Does this web search quality skill work with Gemini 2.5 Flash in chat and paper modes?

Yes, the web search quality skill works with Gemini 2.5 Flash in both chat and paper modes. It applies stage-specific narration and pre-compose source scoring to align responses with research or conversational expectations.

How do I block encyclopedias and content farms from AI search results?

You can block encyclopedias and content farms from AI search results by applying hard-blocked domain filtering. This forces the model to focus exclusively on credible repositories and professional sources during response composition.

What's the best way to ensure AI responses use credible sources?

The best way to ensure AI responses use credible sources is to enforce pre-compose source scoring, credibility heuristics, and reference integrity checks before the model generates its final output. This filters out user-generated hits.

Why does Gemini compose outputs from unverified web sources?

Gemini composes outputs from unverified web sources when pre-compose validation is missing. Without enforced source scoring and reference integrity checks, the model may rely on low-quality, user-generated hits from standard search results.