researching-web
npx machina-cli add skill chris-hendrix/claudehub/researching-web --openclawResearching Web
Methodology for conducting effective web research and synthesizing findings.
Philosophy
- Accuracy first - Verify claims across multiple sources before presenting
- Always cite - Every finding must link back to its source
- Source hierarchy - Official docs > reputable technical sources > community content
- Currency awareness - Note publication dates; flag potentially outdated information
- Multi-angle coverage - Explore topics from multiple search angles for completeness
Critical: Check Current Versions
LLM knowledge is often outdated. When researching libraries, frameworks, or tools:
- Always search for current version - Add "2025" or "latest" to queries
- Check official docs first -
site:docs.{library}.comorsite:{library}.dev - Look for migration guides - New versions often have different patterns
- Verify best practices - What was recommended 2 years ago may be deprecated
Examples of outdated patterns:
- React: Class components → Hooks → Server Components
- ESLint:
.eslintrc→ Flat config (eslint.config.js) - Node.js: CommonJS (
require) → ESM (import) - Next.js: Pages Router → App Router
Always include version information in findings and note when patterns have changed.
Research Types
| Type | Purpose | Tools |
|---|---|---|
| Searching | Find relevant sources | WebSearch |
| Synthesizing | Extract and organize findings | WebFetch |
Choose the appropriate type based on what's needed. They are typically combined: search first, then synthesize from the best results.
Workflow Overview
- Query Analysis - Deconstruct request, identify key terms, plan multiple search angles
- Strategic Searches - Broad to specific, use site-specific targeting, vary query phrasing
- Content Retrieval - Fetch most promising results, prioritize authoritative sources
- Synthesis - Organize by relevance, cite sources, highlight conflicts, note gaps
See searching-web for steps 1-2 and synthesizing-page for steps 3-4.
For query-type-specific strategies, see query-strategies.
Output Structure
Structure findings by purpose:
| Section | Content |
|---|---|
| Summary | 2-3 sentence key takeaways |
| Detailed Findings | Topic sections with source attribution |
| Sources | Links with descriptions |
| Gaps/Limitations | Unresolved questions, outdated info |
Always include source links. Flag conflicting information explicitly.
Source
git clone https://github.com/chris-hendrix/claudehub/blob/main/plugins/rpi/skills/researching-web/SKILL.mdView on GitHub Overview
Researching-web guides systematic online investigation: plan searches, fetch authoritative sources, and synthesize findings with clear citations. It emphasizes accuracy, currency, and a clear source hierarchy to ensure results link back to official docs and reputable sources.
How This Skill Works
The skill uses WebSearch to identify promising sources, then WebFetch to extract and organize content. Findings are synthesized by relevance, with explicit source attribution, notes on currency, and flags for conflicts or outdated information.
When to Use It
- When you need the latest version or migration guides for a library or tool
- When you must locate official docs and site-specific guidance (e.g., site:docs.library.com)
- When resolving a technical issue by cross-checking multiple reputable sources
- When compiling implementation patterns and best practices updated to current versions
- When curating a cited knowledge base with diverse external sources
Quick Start
- Step 1: Analyze the query to extract key terms and version considerations
- Step 2: Run strategic searches using WebSearch, including 'latest'/'2025' and site-specific targets
- Step 3: Use WebFetch to retrieve content, synthesize findings, and list sources with notes on any gaps or conflicts
Best Practices
- Always search for the current version (add 2025 or latest to queries)
- Check official docs first and use site-specific targeting
- Cite every finding with direct links to sources
- Note publication dates and flag potentially outdated information
- Organize results by relevance and clearly document any conflicts or gaps
Example Use Cases
- Researching React 2025 updates by collecting official docs and migration notes with citations
- Tracing Node.js CommonJS to ESM across versions using authoritative sources
- Investigating Next.js routing changes across versions with cited sources
- Summarizing an AWS API docs update and auth changes with sources and version context
- Verifying ESLint config transitions from .eslintrc to eslint.config.js with references
