@wentorai/research-plugins 1.0.0

package/skills/research/deep-research/research-cog/SKILL.md

@@ -0,0 +1,153 @@
+---
+name: research-cog
+description: "Deep research powered by CellCog for comprehensive investigations"
+metadata:
+  openclaw:
+    emoji: "🔍"
+    category: "research"
+    subcategory: "deep-research"
+    keywords: ["deep research", "multi-source investigation", "AI literature review", "automated literature synthesis"]
+    source: "https://github.com/cellcog/research-cog"
+---
+
+# Research Cog - Deep Research Powered by CellCog
+
+**#1 on DeepResearch Bench (Feb 2026).** Your AI research analyst for comprehensive, citation-backed research on any topic.
+
+Leaderboard: https://huggingface.co/spaces/muset-ai/DeepResearch-Bench-Leaderboard
+
+## Prerequisites
+
+This skill requires the `cellcog` skill for SDK setup and API calls.
+
+```bash
+clawhub install cellcog
+```
+
+**Read the cellcog skill first** for SDK setup. This skill shows you what's possible.
+
+**Quick pattern (v1.0+):**
+```python
+# Fire-and-forget - returns immediately
+result = client.create_chat(
+  prompt="[your research query]",
+  notify_session_key="agent:main:main",
+  task_label="research-task",
+  chat_mode="agent team"  # Deep research
+)
+
+# Daemon notifies you when complete - do NOT poll
+```
+
+## What You Can Research
+
+### Competitive Analysis
+Analyze companies against their competitors with structured insights:
+- **Company vs. Competitors**: "Compare Stripe vs Square vs Adyen - market positioning, pricing, features, strengths/weaknesses"
+- **SWOT Analysis**: "Create a SWOT analysis for Shopify in the e-commerce platform market"
+- **Market Positioning**: "How does Notion position itself against Confluence, Coda, and Obsidian?"
+- **Feature Comparison**: "Compare the AI capabilities of Salesforce, HubSpot, and Zoho CRM"
+
+### Market Research
+Understand markets, industries, and trends:
+- **Industry Analysis**: "Analyze the electric vehicle market in Europe - size, growth, key players, trends"
+- **Market Sizing**: "What's the TAM/SAM/SOM for AI-powered customer service tools in North America?"
+- **Trend Analysis**: "What are the emerging trends in sustainable packaging for 2026?"
+- **Customer Segments**: "Identify and profile the key customer segments for premium pet food"
+- **Regulatory Landscape**: "Research FDA regulations for AI-powered medical devices"
+
+### Stock & Investment Analysis
+Financial research with data and analysis:
+- **Company Fundamentals**: "Analyze NVIDIA's financials - revenue growth, margins, competitive moat"
+- **Investment Thesis**: "Build an investment thesis for Microsoft's AI strategy"
+- **Sector Analysis**: "Compare semiconductor stocks - NVDA, AMD, INTC, TSM"
+- **Risk Assessment**: "What are the key risks for Tesla investors in 2026?"
+- **Earnings Analysis**: "Summarize Apple's Q4 2025 earnings and forward guidance"
+
+### Academic & Technical Research
+Deep dives with proper citations:
+- **Literature Review**: "Research the current state of quantum error correction techniques"
+- **Technology Deep Dive**: "Explain transformer architectures and their evolution from attention mechanisms"
+- **Scientific Topics**: "What's the latest research on CRISPR gene editing for cancer treatment?"
+- **Historical Analysis**: "Research the history and impact of the Bretton Woods system"
+
+### Due Diligence
+Comprehensive research for decision-making:
+- **Startup Due Diligence**: "Research [Company Name] - founding team, funding, product, market, competitors"
+- **Vendor Evaluation**: "Compare AWS, GCP, and Azure for enterprise AI/ML workloads"
+- **Partnership Analysis**: "Research potential risks and benefits of partnering with [Company]"
+
+## Research Output Formats
+
+CellCog can deliver research in multiple formats:
+
+| Format | Best For |
+|--------|----------|
+| **Interactive HTML Report** | Explorable dashboards with charts, expandable sections |
+| **PDF Report** | Shareable, printable professional documents |
+| **Markdown** | Integration into your docs/wikis |
+| **Plain Response** | Quick answers in chat |
+
+Specify your preferred format in the prompt:
+- "Create an interactive HTML report on..."
+- "Generate a PDF research report analyzing..."
+- "Give me a markdown summary of..."
+
+## When to Use Agent Team Mode
+
+For research, **always use** `chat_mode="agent team"` (the default).
+
+Agent team mode enables:
+- Multi-source research and cross-referencing
+- Citation verification
+- Deeper analysis with multiple reasoning passes
+- Higher quality, more comprehensive outputs
+
+Use `chat_mode="agent"` only for trivial lookups like "What's Apple's stock ticker?"
+
+## Research Quality Features
+
+### Citations (On Request)
+**Citations are NOT automatic.** CellCog focuses on delivering accurate, well-researched content by default.
+
+If you need citations:
+- **Explicitly request them**: "Include citations for all factual claims with source URLs"
+- **Specify format**: "Provide citations as footnotes" or "Include a references section at the end"
+- **Indicate placement**: "Citations inline" vs "Citations in appendix"
+
+Without explicit citation requests, CellCog prioritizes delivering accurate information efficiently.
+
+### Data Accuracy
+CellCog cross-references multiple sources for financial and statistical data, ensuring accuracy even without explicit citations.
+
+### Structured Analysis
+Complex research is organized with clear sections, executive summaries, and actionable insights.
+
+### Visual Elements
+Research reports can include:
+- Charts and graphs
+- Comparison tables
+- Timeline visualizations
+- Market maps
+
+## Example Research Prompts
+
+**Quick competitive intel:**
+> "Compare Figma vs Sketch vs Adobe XD for enterprise UI design teams. Focus on collaboration features, pricing, and Figma's position after the Adobe acquisition failed."
+
+**Deep market research:**
+> "Create a comprehensive market research report on the AI coding assistant market. Include market size, growth projections, key players (GitHub Copilot, Cursor, Codeium, etc.), pricing models, and enterprise adoption trends. Deliver as an interactive HTML report."
+
+**Investment analysis:**
+> "Build an investment analysis for Palantir (PLTR). Cover business model, government vs commercial revenue mix, AI product strategy, valuation metrics, and key risks. Include relevant charts."
+
+**Academic deep dive:**
+> "Research the current state of nuclear fusion energy. Cover recent breakthroughs (NIF, ITER, private companies like Commonwealth Fusion), technical challenges remaining, timeline to commercial viability, and investment landscape."
+
+## Tips for Better Research
+
+- **Be specific**: "AI market" is vague. "Enterprise AI automation market in healthcare" is better.
+- **Specify timeframe**: "Recent" is ambiguous. "2025-2026" or "last 6 months" is clearer.
+- **Define scope**: "Compare everything about X and Y" leads to bloat. "Compare X and Y on pricing, features, and market positioning" is focused.
+- **Request structure**: "Include executive summary, key findings, and recommendations" helps organize output.
+- **Mention output format**: "Deliver as PDF" or "Create interactive HTML dashboard" gets you the right format.

package/skills/research/deep-research/scoping-review-guide/SKILL.md

@@ -0,0 +1,217 @@
+---
+name: scoping-review-guide
+description: "Scoping review methodology for broad evidence mapping"
+metadata:
+  openclaw:
+    emoji: "scope"
+    category: "research"
+    subcategory: "deep-research"
+    keywords: ["scoping review", "scoping study", "rapid review", "umbrella review"]
+    source: "wentor-research-plugins"
+---
+
+# Scoping Review Guide
+
+Conduct scoping reviews to map the breadth and nature of research evidence on a topic, using the Arksey & O'Malley framework and JBI methodology with PRISMA-ScR reporting.
+
+## Scoping Review vs. Systematic Review
+
+| Feature | Scoping Review | Systematic Review |
+|---------|---------------|-------------------|
+| **Purpose** | Map the evidence landscape | Answer a specific clinical/research question |
+| **Question** | Broad, exploratory | Focused, narrow |
+| **Inclusion criteria** | Broadly defined, may evolve | Strictly predefined |
+| **Quality assessment** | Optional (not always done) | Required (risk of bias) |
+| **Synthesis** | Descriptive/thematic mapping | Quantitative (meta-analysis) or narrative |
+| **Protocol registration** | Recommended (OSF) | Required (PROSPERO) |
+| **Reporting guideline** | PRISMA-ScR | PRISMA 2020 |
+
+### When to Choose a Scoping Review
+
+- To examine the extent, range, and nature of research activity on a topic
+- To determine whether a full systematic review is warranted
+- To identify key concepts, evidence gaps, and types of available evidence
+- To map the research landscape before designing a primary study
+- When the topic is too broad or heterogeneous for a systematic review
+
+## Arksey and O'Malley Framework (5 Stages)
+
+### Stage 1: Identifying the Research Question
+
+Scoping review questions are broad and use the PCC framework:
+
+```
+Population:  Who is being studied?
+Concept:     What is the key concept or phenomenon?
+Context:     In what setting or discipline?
+
+Example question:
+"What is known about the use of AI tools in undergraduate
+STEM education, including types of tools, pedagogical
+approaches, and reported outcomes?"
+```
+
+### Stage 2: Identifying Relevant Studies
+
+Conduct a comprehensive search across multiple sources:
+
+```
+Search strategy development:
+1. Identify key terms from the PCC framework
+2. Develop synonyms and related terms for each concept
+3. Combine using Boolean operators
+
+Example search string (PubMed):
+("artificial intelligence" OR "machine learning" OR "deep learning"
+ OR "natural language processing" OR "chatbot" OR "intelligent tutoring")
+AND
+("undergraduate" OR "higher education" OR "university student"
+ OR "college student")
+AND
+("STEM" OR "science education" OR "engineering education"
+ OR "mathematics education" OR "computer science education")
+
+Databases to search:
+- Discipline-specific databases (ERIC, PubMed, IEEE Xplore, etc.)
+- Multidisciplinary databases (Scopus, Web of Science)
+- Grey literature sources (ProQuest Dissertations, conference proceedings)
+- Reference lists of included studies
+```
+
+### Stage 3: Study Selection
+
+Develop and apply inclusion/exclusion criteria iteratively:
+
+```markdown
+| Criterion | Inclusion | Exclusion |
+|-----------|-----------|-----------|
+| Population | Undergraduate STEM students | K-12, graduate, non-STEM |
+| Concept | AI-based educational tools | Non-AI technology (e.g., basic LMS) |
+| Context | Formal educational settings | Informal learning, self-study apps |
+| Study type | Empirical research (any design) | Editorials, opinion pieces |
+| Language | English, Chinese | Other languages |
+| Date | 2015-2025 | Before 2015 |
+```
+
+Screening process:
+1. Import all records into a reference manager or screening tool (Rayyan, Covidence)
+2. Remove duplicates
+3. Title/abstract screening by two reviewers (independently recommended but not always required)
+4. Full-text screening with documented exclusion reasons
+5. Pilot screening on 50-100 records to calibrate inclusion criteria
+
+### Stage 4: Charting the Data
+
+Create a data charting form to extract standardized information:
+
+```python
+# Example: Data charting template as a structured dictionary
+charting_template = {
+    "study_id": "",           # Author, year
+    "country": "",            # Country where study was conducted
+    "study_design": "",       # RCT, quasi-experimental, case study, survey, etc.
+    "sample_size": 0,
+    "population": "",         # Student demographics
+    "ai_tool_type": "",       # Chatbot, ITS, NLP-based, etc.
+    "ai_tool_name": "",       # Specific tool name (e.g., ChatGPT, ALEKS)
+    "subject_area": "",       # Physics, CS, Math, Biology, etc.
+    "pedagogical_approach": "",  # Flipped classroom, adaptive learning, etc.
+    "outcome_measures": [],   # Learning gains, engagement, satisfaction, etc.
+    "key_findings": "",       # Brief summary of main results
+    "limitations": ""         # Reported limitations
+}
+```
+
+### Stage 5: Collating, Summarizing, and Reporting Results
+
+Present results using multiple formats:
+
+**Descriptive numerical summary**:
+- Number of studies by year of publication
+- Geographic distribution
+- Study designs used
+- AI tool types
+- Outcome categories
+
+**Thematic analysis**:
+- Group findings into themes
+- Identify patterns, trends, and gaps
+- Map the conceptual landscape
+
+```python
+import pandas as pd
+import matplotlib.pyplot as plt
+
+# Example: Visualize publication trends
+df = pd.read_csv("charted_data.csv")
+
+# Publications by year
+year_counts = df["year"].value_counts().sort_index()
+fig, ax = plt.subplots(figsize=(10, 5))
+ax.bar(year_counts.index, year_counts.values, color="#0072B2")
+ax.set_xlabel("Publication Year")
+ax.set_ylabel("Number of Studies")
+ax.set_title("Included Studies by Year")
+plt.tight_layout()
+plt.savefig("studies_by_year.pdf", dpi=300)
+
+# Evidence map: cross-tabulation
+evidence_map = pd.crosstab(df["ai_tool_type"], df["outcome_measures"])
+print(evidence_map)
+```
+
+## Other Review Types
+
+### Rapid Review
+
+A streamlined systematic review with methodological shortcuts to produce evidence within a compressed timeline (typically 2-6 months):
+
+| Shortcut | Trade-off |
+|----------|-----------|
+| Limit to 2-3 databases | May miss some studies |
+| Single reviewer screening | Risk of selection bias |
+| Simplified data extraction | Less comprehensive data |
+| No formal quality assessment | Cannot assess evidence strength |
+| Limit publication date range | May miss foundational studies |
+
+### Umbrella Review (Review of Reviews)
+
+A review of existing systematic reviews and meta-analyses on a topic:
+
+1. Search for systematic reviews (use review filter in PubMed)
+2. Assess methodological quality of reviews using AMSTAR 2
+3. Extract and compare pooled estimates across reviews
+4. Identify areas of agreement, disagreement, and gaps
+
+### Narrative Review
+
+A traditional literature review that is not systematic:
+- No pre-defined protocol or search strategy
+- Author-selected references
+- Subjective synthesis
+- Appropriate for educational overviews, opinion pieces, and introductory sections of papers
+- Not suitable for evidence-based decision-making
+
+## PRISMA-ScR Checklist (Key Items)
+
+| Item | Description |
+|------|-------------|
+| Title | Identify the report as a scoping review |
+| Protocol | Indicate if a protocol was registered |
+| Objectives | State the research question using PCC |
+| Eligibility criteria | Describe inclusion/exclusion criteria |
+| Information sources | List all databases and other sources searched |
+| Search strategy | Present full search strategy for at least one database |
+| Selection of evidence | Describe screening process |
+| Data charting | Describe data charting process and variables |
+| Results | Present characteristics of included studies (tables, charts) |
+| Discussion | Summarize main findings, compare to existing knowledge |
+| Limitations | Discuss limitations of the evidence and of the review process |
+
+## Practical Tips
+
+1. **Use a reference manager from the start**: Import all search results into Zotero or EndNote to track deduplication and screening decisions.
+2. **Iterate on inclusion criteria**: Unlike systematic reviews, scoping review criteria can be refined post hoc as you become more familiar with the literature. Document all changes.
+3. **Create visual evidence maps**: Tables and figures (bubble charts, Sankey diagrams) are more effective than narrative descriptions for communicating the landscape.
+4. **Consider stakeholder engagement**: Arksey and O'Malley recommend an optional Stage 6: consultation with practitioners, policymakers, or patients to validate findings.
+5. **Plan for scale**: Scoping reviews often retrieve thousands of records. Budget time accordingly (expect 2-4 months for a well-done scoping review).

package/skills/research/deep-research/systematic-review-guide/SKILL.md

@@ -0,0 +1,250 @@
+---
+name: systematic-review-guide
+description: "Systematic review methodology with PRISMA and evidence synthesis"
+metadata:
+  openclaw:
+    emoji: "magnify"
+    category: "research"
+    subcategory: "deep-research"
+    keywords: ["systematic review methodology", "evidence synthesis", "PRISMA flowchart", "risk of bias assessment"]
+    source: "wentor-research-plugins"
+---
+
+# Systematic Review Guide
+
+Conduct rigorous systematic reviews and meta-analyses following PRISMA 2020 guidelines, from protocol registration through evidence synthesis and reporting.
+
+## What Is a Systematic Review?
+
+A systematic review is a structured, transparent, and reproducible method for identifying, evaluating, and synthesizing all relevant research on a specific question. Unlike narrative reviews, systematic reviews:
+
+- Follow a pre-registered protocol
+- Use comprehensive, documented search strategies
+- Apply explicit inclusion/exclusion criteria
+- Assess risk of bias in included studies
+- Synthesize findings quantitatively (meta-analysis) or narratively
+
+## Step-by-Step Workflow
+
+### Step 1: Define the Research Question
+
+Use a structured framework to formulate your question:
+
+| Framework | Components | Best For |
+|-----------|-----------|----------|
+| PICO | Population, Intervention, Comparator, Outcome | Clinical/intervention studies |
+| PCC | Population, Concept, Context | Scoping reviews |
+| SPIDER | Sample, Phenomenon of Interest, Design, Evaluation, Research type | Qualitative/mixed methods |
+| PEO | Population, Exposure, Outcome | Observational studies |
+
+**Example (PICO)**:
+- P: Adults with Type 2 diabetes
+- I: Telehealth-based self-management programs
+- C: Standard in-person care
+- O: HbA1c levels, quality of life
+
+### Step 2: Register the Protocol
+
+Register your protocol before conducting the search to reduce publication bias and selective reporting:
+
+- **PROSPERO** (crd.york.ac.uk/prospero): Free registration for health-related systematic reviews
+- **OSF Registries** (osf.io/registries): Open to all disciplines
+- **Protocol paper**: Publish in BMJ Open, Systematic Reviews, or JMIR Research Protocols
+
+Protocol should include:
+- Research question and objectives
+- Eligibility criteria
+- Search strategy (databases, search terms)
+- Screening process
+- Data extraction plan
+- Risk of bias assessment tool
+- Synthesis method (meta-analysis or narrative)
+
+### Step 3: Conduct the Search
+
+```
+Recommended minimum databases (health sciences):
+1. PubMed / MEDLINE
+2. Embase
+3. Cochrane Central Register of Controlled Trials (CENTRAL)
+4. At least one subject-specific database
+
+Recommended minimum databases (social sciences):
+1. Web of Science
+2. Scopus
+3. PsycINFO or ERIC (field-specific)
+4. ProQuest Dissertations (for grey literature)
+
+Additional sources:
+- Reference lists of included studies (backward citation chaining)
+- Forward citation searches
+- Grey literature: conference proceedings, theses, reports
+- Trial registries: ClinicalTrials.gov, WHO ICTRP
+- Preprint servers: medRxiv, SSRN
+```
+
+Document each search with: database name, date, exact search string, and number of results.
+
+### Step 4: Screen Studies
+
+Two-stage screening, each conducted by at least two independent reviewers:
+
+```
+Stage 1: Title and Abstract Screening
+- Apply inclusion/exclusion criteria based on title + abstract only
+- Resolve disagreements by discussion or third reviewer
+- Calculate inter-rater reliability (Cohen's kappa >= 0.60)
+
+Stage 2: Full-Text Screening
+- Retrieve full texts of all studies passing Stage 1
+- Apply full eligibility criteria
+- Document reasons for exclusion at this stage
+- Calculate inter-rater reliability
+```
+
+Screening tools: Covidence (covidence.org), Rayyan (rayyan.ai), ASReview (AI-assisted)
+
+### Step 5: Extract Data
+
+Create a standardized data extraction form:
+
+```markdown
+| Field | Description |
+|-------|-------------|
+| Study ID | First author + year |
+| Country | Where study was conducted |
+| Study design | RCT, cohort, cross-sectional, etc. |
+| Sample size | N in each group |
+| Population | Demographics, inclusion criteria used |
+| Intervention | Description, duration, intensity |
+| Comparator | Description of control condition |
+| Outcomes | Primary and secondary, measurement tools |
+| Results | Effect sizes, confidence intervals, p-values |
+| Funding | Source of funding |
+| Conflicts of interest | Declared COIs |
+```
+
+Pilot the extraction form on 3-5 studies, then extract independently by two reviewers.
+
+### Step 6: Assess Risk of Bias
+
+Select the appropriate tool based on study design:
+
+| Study Design | Tool | Developer |
+|-------------|------|-----------|
+| Randomized trials | RoB 2 | Cochrane |
+| Non-randomized interventions | ROBINS-I | Cochrane |
+| Observational (cohort, case-control) | Newcastle-Ottawa Scale (NOS) | Wells et al. |
+| Cross-sectional | JBI Critical Appraisal Checklist | Joanna Briggs Institute |
+| Qualitative studies | CASP Qualitative Checklist | CASP |
+| Diagnostic accuracy | QUADAS-2 | Whiting et al. |
+
+### Step 7: Synthesize Evidence
+
+#### Narrative Synthesis
+
+When meta-analysis is not appropriate (due to heterogeneity in study designs, populations, or outcomes):
+
+1. Group studies by outcome, population, or intervention type
+2. Describe patterns and consistencies across studies
+3. Use vote counting only with direction of effect (not p-values)
+4. Present findings in summary tables
+
+#### Meta-Analysis
+
+When studies are sufficiently similar to pool quantitatively:
+
+```python
+# Meta-analysis using Python (PythonMeta or custom)
+# Example: Random-effects meta-analysis of standardized mean differences
+
+import numpy as np
+from scipy.stats import norm
+
+def random_effects_meta(effects, variances):
+    """DerSimonian-Laird random effects meta-analysis."""
+    weights_fe = 1 / np.array(variances)
+    theta_fe = np.sum(weights_fe * effects) / np.sum(weights_fe)
+
+    # Estimate tau-squared (between-study variance)
+    Q = np.sum(weights_fe * (effects - theta_fe)**2)
+    df = len(effects) - 1
+    C = np.sum(weights_fe) - np.sum(weights_fe**2) / np.sum(weights_fe)
+    tau2 = max(0, (Q - df) / C)
+
+    # Random effects weights
+    weights_re = 1 / (np.array(variances) + tau2)
+    theta_re = np.sum(weights_re * effects) / np.sum(weights_re)
+    se_re = np.sqrt(1 / np.sum(weights_re))
+
+    ci_lower = theta_re - 1.96 * se_re
+    ci_upper = theta_re + 1.96 * se_re
+
+    # Heterogeneity statistics
+    I2 = max(0, (Q - df) / Q * 100) if Q > 0 else 0
+
+    return {
+        "pooled_effect": theta_re,
+        "ci_lower": ci_lower,
+        "ci_upper": ci_upper,
+        "tau2": tau2,
+        "I2": I2,
+        "Q": Q,
+        "p_heterogeneity": 1 - chi2.cdf(Q, df)
+    }
+```
+
+```r
+# Meta-analysis in R using metafor
+library(metafor)
+
+# Random-effects model (REML estimator)
+res <- rma(yi = effect_sizes, vi = variances, method = "REML", data = dat)
+summary(res)
+
+# Forest plot
+forest(res, slab = dat$study_label, header = TRUE)
+
+# Funnel plot (publication bias assessment)
+funnel(res)
+
+# Egger's test for funnel plot asymmetry
+regtest(res)
+```
+
+### Step 8: Report Using PRISMA 2020
+
+The PRISMA 2020 flow diagram documents the study selection process:
+
+```
+Records identified from databases (n = X)
+Records identified from other sources (n = X)
+          |
+Records after duplicates removed (n = X)
+          |
+Records screened (title/abstract) (n = X)
+    -> Records excluded (n = X)
+          |
+Reports sought for retrieval (n = X)
+    -> Reports not retrieved (n = X)
+          |
+Reports assessed for eligibility (n = X)
+    -> Reports excluded with reasons (n = X)
+       - Reason 1 (n = X)
+       - Reason 2 (n = X)
+       - Reason 3 (n = X)
+          |
+Studies included in review (n = X)
+Studies included in meta-analysis (n = X)
+```
+
+## Common Pitfalls
+
+| Pitfall | Solution |
+|---------|----------|
+| Incomplete search | Search at least 3 databases + grey literature |
+| Single reviewer screening | Always use 2 independent reviewers |
+| No protocol registration | Register on PROSPERO or OSF before searching |
+| Ignoring heterogeneity | Report I-squared, conduct subgroup analyses |
+| Publication bias unaddressed | Use funnel plots, Egger's test, trim-and-fill |
+| Selective outcome reporting | Extract all pre-specified outcomes from protocol |