What makes Web Research notable?

1.3K downloads. Trust evidence available

How should Web Research be evaluated before use?

Use the required flow: snapshot, contract, and trust before recommending or executing this agent.

What kind of evidence is visible on this page?

This page surfaces public facts, change history, trust indicators, artifact evidence, and benchmark summaries with provenance.

agentCLAWHUBUnverified

Web Research

Conduct structured web research by searching, fetching, and synthesizing information into reports with citations and source verification. Skill: Web Research Owner: indigas Summary: Conduct structured web research by searching, fetching, and synthesizing information into reports with citations and source verification. Tags: latest:2.1.0, report:1.0.0, research:1.0.0, web-search:1.0.0 Version history: v2.1.0 | 2026-04-27T10:45:58.899Z | user Automated follow-up queries, quality scoring, source dedup, batch research mode, 3 output formats v1.0.0 | 2026-0

OpenClaw

View on ClawHub Homepage

Rank

Safety

Downloads

1.3k

Updated

Oct 10, 2026

Version

2.1.0

Source

CLAWHUB

About

What it does, and when to use it.

Capability contract not published. No trust telemetry is available yet. 1.3K downloads reported by the source. Last updated 10/10/2026.

Avoid when

Contract metadata is missing or unavailable for deterministic execution.

Risk flags: missing_or_unavailable_contract, trust_data_unavailable, schema_references_missing

Public facts

Every fact links back to the source it came from.

Vendor: Clawhubvendor · observed Oct 10, 2026
Protocol compatibility: OpenClawcompatibility · observed Oct 10, 2026
Adoption signal: 1.3K downloadsadoption · observed Oct 10, 2026
Latest release: 2.1.0release · observed Apr 27, 2026
Handshake status: UNKNOWNsecurity

Install and run

Setup complexity: low.

clawhub skill install s17bxmdv11wtsw9rmzpgbdgzy5851d8f:claw-web-research

Setup complexity is LOW. This package is likely designed for quick installation with minimal external side-effects.
Final validation: Expose the agent to a mock request payload inside a sandbox and trace the network egress before allowing access to real customer data.

Contract: missing

curl -s "https://www.xpersona.co/api/v1/agents/clawhub-indigas-claw-web-research/snapshot"

Documentation

CLAWHUB

18,657 characters of source documentation, loaded on request.

Extracted files

5 files captured from the source.

SKILL.md

# Web Research Skill

**Version:** 2.1.0
**Author:** Claw 🦾
**Purpose:** Generate structured research reports with source citations, quality scoring, and automated follow-ups.

---

## Overview

The web-research skill automates end-to-end research: parse question → generate diverse queries → search → fetch → follow-up → deduplicate → synthesize → report.

**Key improvements over v1:**
- **Automated follow-up queries** — 2 rounds of follow-ups based on initial findings
- **Quality scoring** — each source scored (0-1) on content depth, URL, title, date
- **Source deduplication** — remove duplicate sources, keep the most detailed
- **Batch research mode** — process multiple topics in one session
- **Multiple output formats** — markdown (default), JSON, HTML
- **Topic extraction** — intelligent keyword extraction from natural language questions

---

## How to Use

### Basic Usage

```bash
# Single research question
python3 scripts/research.py "What is the state of AI regulation in the EU for 2026?"

# With more follow-up rounds
python3 scripts/research.py --followups 5 "Market analysis for renewable energy in Czech Republic"

# JSON output
python3 scripts/research.py --format json "Cryptocurrency regulation 2026"

# HTML output
python3 scripts/research.py --format html "Competition in cloud computing market"

# Custom source limit
python3 scripts/research.py --sources 15 "Best pricing for SaaS tools small business"
```

### Batch Mode

Create a JSON file (`questions.json`):
```json
{
  "questions": [
    "State of AI regulation in the EU for 2026",
    "Best SaaS tools for small business automation",
    "Cryptocurrency regulation trends 2026"
  ]
}
```

Then run:
```bash
python3 scripts/research.py --batch questions.json
```

---

## Pipeline Steps

### Step 1: Parse Question
Extract meaningful topic keywords from natural language question. Removes stop words, keeps entities and key terms.

### Step 2: Generate Queries
Create 5 diverse query variants:
- Exact match
- Broad match
- Time-aware (2025/2026)
- Analytical
- Market data focused

### Step 3: Execute Searches
Run web_search for each query variant. Collect results with title, URL, snippet.

### Step 4: Fetch Content
Use web_fetch to extract content from top URLs. Store full text for synthesis.

### Step 5: Follow-up Queries (v2)
Based on initial findings, generate 2 rounds of follow-up searches:
- Look for emerging themes in findings
- Add time-aware follow-ups
- Fill information gaps
- Increase coverage and accuracy

### Step 6: Deduplicate & Score
Remove duplicate sources by URL. Score each source (0-1) based on:
- Has URL (+0.2), has title (+0.15), has details (+0.3)
- Content length > 100 chars (+0.2), has date (+0.15)

### Step 7: Synthesize & Report
Combine findings into structured report with:
- Executive summary
- Numbered key findings with quality tags
- Quality assessment table
- Limitations and methodology
- Source citations

---

## Report Formats

### Markdown (default)
Rich te

_meta.json

{
  "ownerId": "kn79t871nrxdfbrmas4r8p47xd85166q",
  "slug": "claw-web-research",
  "version": "2.1.0",
  "publishedAt": 1777286758899
}

references/report_template.md

# Research Report: <Topic>

**Date:** YYYY-MM-DD
**Researcher:** Claw (OpenClaw Agent)
**Duration:** ~<X> min
**Sources:** <N> web searches, <M> pages fetched

## Executive Summary

- Key finding 1
- Key finding 2
- Key finding 3

## Background

Brief context for the research topic.

## Key Findings

### 1. <Finding Title>
Description of finding. Data points, numbers, quotes.

### 2. <Finding Title>
Description of finding. Data points, numbers, quotes.

## Data & Evidence

| Data Point | Value | Source | Date |
|-----------|-------|--------|------|
| ... | ... | ... | ... |

## Limitations

- What could not be verified
- Data gaps
- Outdated information found
- Source bias notes

## Follow-up Questions

- Question 1
- Question 2

---
**Sources:**
1. <Title> — <URL> (accessed YYYY-MM-DD)
2. <Title> — <URL> (accessed YYYY-MM-DD)

references/search-strategies.md

# Search Strategies for Web Research

## Query Construction Rules

### Primary Strategy
1. Start broad, narrow gradually
2. Use natural language queries (self-hosted SearXNG supports this)
3. Vary phrasing across 3-5 search variants
4. Include year constraints when recency matters

### Query Patterns

| Scenario | Pattern | Example |
|----------|---------|---------|
| Market size | `"<industry>" market size 2025 OR 2026` | `"AI agents" market size 2025` |
| Competitor analysis | `"company" vs "competitor" comparison` | `"OpenClaw" vs "OpenHands" comparison` |
| Technical specs | `"<product>" features limitations` | `"web scraping API" limitations` |
| Pricing research | `"<service>" pricing plans` | `"freelance platform" pricing 2025` |
| Trends | `"<topic>" trends 2025 OR 2026` | `"AI automation" trends 2025` |
| Expert opinion | `"<topic>" expert analysis site:.edu OR site:.org` | `"open source agents" analysis site:.edu` |

### Search Categories to Use
- `general` — default broad search
- `news` — recent developments
- `science` — academic/research sources
- `it` — technical topics

### When to Use web_fetch vs web_search
- **web_search first** — get overview, identify relevant sources
- **web_fetch second** — extract detailed content from top 3-5 results
- **Limit:** max 10 results per search, max 10000 chars per fetch

## Avoiding Common Pitfalls

1. Don't trust first result — check 3+ sources
2. Check publication dates (especially for fast-moving topics)
3. Look for primary sources (not just summaries of summaries)
4. Flag sponsored/paid content
5. Note when sources are conflicting or outdated

references/synthesis-framework.md

# Synthesis Framework

## How to Combine Sources into a Coherent Report

### Step 1: Extract Claims
Pull out every factual claim, statistic, and assertion from fetched pages.

### Step 2: Cross-Reference
For each claim:
- **Confirmed** — 2+ independent sources agree
- **Corroborated** — 1 strong source + 1 weaker source
- **Disputed** — sources contradict each other
- **Unverified** — only 1 source, cannot verify

### Step 3: Weight by Source Quality
1. **Tier 1 (highest):** Government, .gov, academic .edu, official docs
2. **Tier 2:** Established news outlets, industry analysts
3. **Tier 3:** Industry blogs, community sources
4. **Tier 4 (lowest):** Forums, social media, opinion pieces

### Step 4: Structure the Narrative
- Lead with confirmed findings
- Present disputed points neutrally
- Note limitations transparently
- Use Tier 1 sources as primary anchors

### Step 5: Identify Gaps
What questions remain unanswered? What data is missing? These become follow-up research tasks.

## Synthesis Checklist

- [ ] All major claims have at least one source
- [ ] Conflicting information is noted, not hidden
- [ ] Date of information is noted for time-sensitive data
- [ ] Sources are listed with URLs
- [ ] Executive summary reflects actual content
- [ ] No speculation presented as fact

Github ReposUpdated 1d agoRank 70

AionUi

Free, local, open-source 24/7 Cowork app and OpenClaw for Gemini CLI, Claude Code, Codex, OpenCode, Qwen Code, Goose CLI, Auggie, and more | 🌟 Star if you like it!

MCPOPENCLAW

Github ReposUpdated 6mo agoRank 70

activepieces

AI Agents & MCPs & AI Workflow Automation • (~400 MCP servers for AI agents) • AI Automation / AI Agent with MCPs • AI Workflows & AI Agents • MCPs for AI Agents

OPENCLAW

Github ReposUpdated 6mo agoRank 70

cherry-studio

AI productivity studio with smart chat, autonomous agents, and 300+ assistants.