1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
|
{
"name": "advanced-web-research-workflow",
"description": "Professional guidelines for using search, fetch, and extract commands to perform deep-dive research, data mining, and synthesis.",
"tags": [
"research",
"search",
"fetch",
"extraction",
"synthesis",
"intelligence"
],
"category": "general",
"created": "2026-06-23T11:36:58.020Z",
"updated": "2026-06-24T21:58:03.530Z",
"content": "---\nname: advanced-web-research-workflow\ndescription: Professional guidelines for using search, fetch, and extract commands to perform deep-dive research, data mining, and synthesis.\ntags: [research, search, fetch, extraction, synthesis, intelligence]\n---\n\n# Advanced Web Research & Extraction Workflow\n\nThis skill defines the operational standard for transforming raw web data into structured intelligence using the `SearXNG` and `WebFetcher` modules.\n\n## 🛠 Tool Mapping\n| Command | Tool / Module | Primary Purpose | Best Use Case |\n| :--- | :--- | :--- | :--- |\n| `search` / `s` | `SearXNGModule` | Broad Discovery | Finding multiple sources, comparing viewpoints, or identifying key URLs. |\n| `fetch` | `WebFetcherModule` | Raw Data Acquisition | Reading the full content of a specific page for deep analysis. |\n| `extract` | `WebFetcherModule` | Targeted Data Mining | Pulling specific structured data or cleaning noise from a page. |\n\n## 🔄 The Intelligence Loop (S-F-E-S)\n\nTo avoid \"hallucination\" and ensure accuracy, follow the **S-F-E-S** loop:\n\n### 1. Search (Discovery) 🔍\n* **Strategy:** Use specific keywords and operators.\n* **Goal:** Generate a list of 3-5 high-authority URLs.\n* **Execution:** `search \"query topic\"` $\\rightarrow$ Scan titles $\\rightarrow$ Select targets.\n\n### 2. Fetch (Acquisition) 📥\n* **Strategy:** Fetch the most promising links.\n* **Goal:** Load the full text into the AI's active context window.\n* **Execution:** `fetch \"https://url.com\"` $\\rightarrow$ Read content.\n\n### 3. Extract (Distillation) ⚡\n* **Strategy:** Use the `extract` command for long-form pages or noisy HTML.\n* **Goal:** Isolate the \"signal\" (facts, numbers, quotes) from the \"noise\" (ads, footers).\n* **Execution:** `extract \"https://url.com\"` $\\rightarrow$ Focus on core data.\n\n### 4. Synthesize (Intelligence) 🧠\n* **Strategy:** Cross-reference data from multiple fetched sources.\n* **Goal:** Produce a summary, report, or risk matrix.\n* **Execution:** Compare Source A vs Source B $\\rightarrow$ Generate final output.\n\n## ⚠️ Auditor's Guardrails (Verification)\n* **Cross-Verification:** Never rely on a single source for critical facts.\n* **URL Validation:** Check if the domain is a known authority or a blog.\n* **Content Recency:** Check the date of the page to ensure the information isn't deprecated.\n* **Anti-Hallucination:** Always cite the specific URL when providing a fact.\n\n## 🚀 Example Workflow\n**Objective:** Audit the security of a new API.\n1. `search \"API Name security vulnerabilities CVE\"` $\\rightarrow$ Find a list of CVEs.\n2. `fetch \"https://cve.mitre.org/cgi-bin/cvename.cgi?na=CVE-...\"` $\\rightarrow$ Read the specific flaw.\n3. `extract \"https://api-docs.com/security\"` $\\rightarrow$ Get the official security implementation details.\n4. **Synthesize:** Map the CVE flaw against the official implementation to find the gap.\n"
}
|
