@wentorai/research-plugins 1.0.0 → 1.2.0

package/skills/research/deep-research/llm-scientific-discovery-guide/SKILL.md

@@ -0,0 +1,178 @@
+---
+name: llm-scientific-discovery-guide
+description: "Survey of LLM agents for biomedical scientific discovery"
+metadata:
+  openclaw:
+    emoji: "🧬"
+    category: "research"
+    subcategory: "deep-research"
+    keywords: ["LLM agents", "scientific discovery", "biomedical AI", "drug discovery", "hypothesis generation", "lab automation"]
+    source: "https://github.com/zjlrock777/Awesome-LLM-Agents-Scientific-Discovery"
+---
+
+# LLM Agents for Scientific Discovery Guide
+
+## Overview
+
+A curated survey of how LLM-based agents are being applied to scientific discovery, with a focus on biomedical research. Covers hypothesis generation, experiment design, lab automation, literature synthesis, and multi-agent scientific collaboration. Tracks papers, tools, and frameworks across the spectrum from fully autonomous to human-in-the-loop systems.
+
+## Landscape
+
+```
+LLM Agents for Scientific Discovery
+├── Hypothesis Generation
+│   ├── Literature-based (gap identification)
+│   ├── Data-driven (pattern discovery)
+│   └── Analogy-based (cross-domain transfer)
+├── Experiment Design
+│   ├── Protocol generation
+│   ├── Parameter optimization
+│   └── Control selection
+├── Lab Automation
+│   ├── Robot control (self-driving labs)
+│   ├── Equipment programming
+│   └── Data collection orchestration
+├── Analysis & Interpretation
+│   ├── Statistical analysis
+│   ├── Visualization
+│   └── Result interpretation
+└── Communication
+    ├── Paper writing
+    ├── Presentation generation
+    └── Peer review simulation
+```
+
+## Key Systems
+
+| System | Domain | Capability |
+|--------|--------|-----------|
+| **AI Scientist** | ML/AI | Full paper generation pipeline |
+| **ChemCrow** | Chemistry | Tool-augmented chemical reasoning |
+| **Coscientist** | Chemistry | Autonomous experiment execution |
+| **BioPlanner** | Biology | Experiment protocol generation |
+| **MedAgent** | Medicine | Clinical trial analysis |
+| **GenAgent** | Genomics | Gene expression analysis |
+| **DrugAgent** | Pharma | Drug interaction prediction |
+
+## Hypothesis Generation
+
+```python
+# LLM-based hypothesis generation pattern
+from scientific_agent import HypothesisGenerator
+
+generator = HypothesisGenerator(
+    llm_provider="anthropic",
+    knowledge_sources=["pubmed", "semantic_scholar"],
+)
+
+hypotheses = generator.generate(
+    domain="oncology",
+    context="Recent findings show that gut microbiome "
+            "composition correlates with immunotherapy response",
+    constraints=[
+        "Must be testable in vitro",
+        "Should involve specific bacterial species",
+        "Must have measurable endpoints",
+    ],
+    num_hypotheses=5,
+)
+
+for h in hypotheses:
+    print(f"\nHypothesis: {h.statement}")
+    print(f"  Rationale: {h.rationale}")
+    print(f"  Supporting evidence: {len(h.evidence)} papers")
+    print(f"  Novelty score: {h.novelty_score:.2f}")
+    print(f"  Feasibility: {h.feasibility}")
+```
+
+## Self-Driving Lab Integration
+
+```python
+# Agent controlling automated experiments
+from scientific_agent import LabAgent
+
+agent = LabAgent(
+    llm_provider="anthropic",
+    equipment=["plate_reader", "liquid_handler", "incubator"],
+    safety_constraints=["bsl2", "max_volume_1ml"],
+)
+
+# Design and run experiment
+result = agent.run_experiment(
+    objective="Determine IC50 of compound X against cell line Y",
+    protocol_type="dose_response",
+    parameters={
+        "compound": "Compound_X",
+        "cell_line": "HeLa",
+        "concentrations": "serial_dilution",
+        "replicates": 3,
+        "readout": "cell_viability",
+    },
+)
+
+print(f"IC50: {result.ic50:.2f} uM")
+print(f"R-squared: {result.r_squared:.3f}")
+result.plot_dose_response("dose_response.pdf")
+```
+
+## Multi-Agent Scientific Collaboration
+
+```python
+# Agents with different scientific roles
+from scientific_agent import ScientificTeam
+
+team = ScientificTeam(
+    agents={
+        "PI": {"role": "research_director",
+               "expertise": "oncology"},
+        "Experimentalist": {"role": "experiment_design",
+                           "expertise": "cell_biology"},
+        "Analyst": {"role": "data_analysis",
+                   "expertise": "biostatistics"},
+        "Writer": {"role": "manuscript_writing",
+                  "expertise": "scientific_communication"},
+    },
+)
+
+# Collaborative research cycle
+project = team.start_project(
+    title="Microbiome-immunotherapy interaction study",
+    timeline_weeks=12,
+)
+
+# Agents collaborate: PI directs → Experimentalist designs →
+# Analyst processes → Writer documents
+```
+
+## Reading Roadmap
+
+```markdown
+### Foundational Papers
+1. "The AI Scientist" (Lu et al., 2024) — Fully automated ML research
+2. "ChemCrow" (Bran et al., 2023) — Chemistry tool-use agent
+3. "Coscientist" (Boiko et al., 2023) — Autonomous chemical research
+4. "BioPlanner" (Biswas et al., 2024) — Biology protocol generation
+
+### Surveys
+5. "Scientific Discovery in the Age of AI" (Wang et al., 2023)
+6. "Foundation Models for Science" (Bommasani et al., 2022)
+7. "LLM Agents: A Survey" (multiple, 2024)
+
+### Ethics & Limitations
+8. "Dual-use concerns of AI in biology" (Sandbrink, 2023)
+9. "Can LLMs Generate Novel Research Ideas?" (Si et al., 2024)
+```
+
+## Use Cases
+
+1. **Literature mining**: Automated hypothesis from research gaps
+2. **Experiment automation**: Self-driving lab orchestration
+3. **Drug discovery**: Multi-agent screening and optimization
+4. **Research planning**: Protocol and proposal generation
+5. **Scientific writing**: Paper drafting with verified claims
+
+## References
+
+- [Awesome-LLM-Agents-Scientific-Discovery](https://github.com/zjlrock777/Awesome-LLM-Agents-Scientific-Discovery)
+- [The AI Scientist](https://arxiv.org/abs/2408.06292)
+- [ChemCrow](https://arxiv.org/abs/2304.05376)

package/skills/research/deep-research/local-deep-research-guide/SKILL.md

@@ -0,0 +1,253 @@
+---
+name: local-deep-research-guide
+description: "Deep research agent searching 10+ sources with local or cloud LLMs"
+metadata:
+  openclaw:
+    emoji: "🏠"
+    category: "research"
+    subcategory: "deep-research"
+    keywords: ["local-llm", "deep-research", "multi-source", "ollama", "privacy", "academic-search"]
+    source: "https://github.com/LearningCircuit/local-deep-research"
+---
+
+# Local Deep Research Guide
+
+## Overview
+
+Local Deep Research is an open-source deep research tool with over 4,000 GitHub stars that conducts comprehensive multi-source research using either local LLMs (via Ollama, LM Studio, or vLLM) or cloud-based models. It searches across 10+ academic and web sources simultaneously, synthesizes the findings, and produces well-cited research reports. The project is designed for researchers who need thorough, multi-perspective research coverage while maintaining the option to keep everything running locally for privacy.
+
+What makes Local Deep Research stand out is its breadth of search integration. Rather than relying on a single search API, it queries multiple sources in parallel -- including Google Scholar, Semantic Scholar, arXiv, PubMed, Wikipedia, web search engines, and more -- then cross-references and synthesizes the results. This multi-source approach produces more comprehensive and balanced research outputs compared to single-source tools.
+
+The tool is particularly well-suited for academic researchers who need to conduct preliminary literature reviews, verify claims across multiple databases, or explore interdisciplinary topics where relevant work may be scattered across different platforms and publication venues.
+
+## Installation and Setup
+
+```bash
+# Install from PyPI
+pip install local-deep-research
+
+# Or clone for development
+git clone https://github.com/LearningCircuit/local-deep-research.git
+cd local-deep-research
+pip install -e .
+```
+
+### LLM Backend Configuration
+
+Local Deep Research supports multiple LLM backends. Choose the one that fits your privacy and performance requirements:
+
+```bash
+# Option 1: Local LLM via Ollama (fully private)
+# First, install Ollama: https://ollama.com/
+ollama pull llama3.1:70b
+export LDR_LLM_PROVIDER=ollama
+export LDR_LLM_MODEL=llama3.1:70b
+
+# Option 2: Local LLM via LM Studio
+export LDR_LLM_PROVIDER=lmstudio
+export LDR_LLM_BASE_URL=http://localhost:1234/v1
+
+# Option 3: Cloud LLM (OpenAI)
+export LDR_LLM_PROVIDER=openai
+export OPENAI_API_KEY=$OPENAI_API_KEY
+export LDR_LLM_MODEL=gpt-4o
+
+# Option 4: Cloud LLM (Anthropic)
+export LDR_LLM_PROVIDER=anthropic
+export ANTHROPIC_API_KEY=$ANTHROPIC_API_KEY
+export LDR_LLM_MODEL=claude-sonnet-4-20250514
+```
+
+### Search Source Configuration
+
+Configure which search sources to use:
+
+```bash
+# Web search (at least one required)
+export SERPER_API_KEY=$SERPER_API_KEY
+# Or
+export TAVILY_API_KEY=$TAVILY_API_KEY
+# Or
+export SEARX_URL=http://localhost:8888  # Self-hosted SearXNG
+
+# Academic sources (optional, enhances academic research)
+export SEMANTIC_SCHOLAR_API_KEY=$SEMANTIC_SCHOLAR_API_KEY
+# PubMed and arXiv require no API keys
+```
+
+## Core Research Capabilities
+
+### Running a Research Query
+
+Start a research session from the command line or Python API:
+
+```bash
+# Command-line interface
+local-deep-research "What are the most effective methods for \
+  few-shot learning in NLP as of 2024?"
+```
+
+```python
+# Python API
+from local_deep_research import DeepResearcher
+
+researcher = DeepResearcher(
+    llm_provider="ollama",
+    llm_model="llama3.1:70b",
+    search_sources=["google_scholar", "semantic_scholar",
+                    "arxiv", "web"],
+    max_iterations=10,
+)
+
+result = researcher.research(
+    "What are the most effective methods for few-shot learning "
+    "in NLP as of 2024?"
+)
+
+print(result.report)
+```
+
+### Multi-Source Search Engine
+
+Local Deep Research queries multiple sources in parallel for each research sub-question:
+
+| Source | Type | API Key Required | Best For |
+|--------|------|-----------------|----------|
+| Google Scholar | Academic | No (via scraping) | Broad academic search |
+| Semantic Scholar | Academic | Optional | CS/AI papers, citation data |
+| arXiv | Academic | No | Preprints, ML/physics/math |
+| PubMed | Academic | No | Biomedical literature |
+| Wikipedia | Encyclopedia | No | Background and definitions |
+| Web Search | General | Yes (Serper/Tavily) | Recent developments |
+| SearXNG | Meta-search | Self-hosted | Privacy-focused web search |
+| CrossRef | Academic | No | DOI resolution, metadata |
+| CORE | Academic | Optional | Open access papers |
+| Unpaywall | Academic | No | Open access PDF links |
+
+```python
+# Customize source priorities for your research domain
+researcher = DeepResearcher(
+    search_sources={
+        "primary": ["semantic_scholar", "arxiv"],
+        "secondary": ["google_scholar", "web"],
+        "reference": ["wikipedia", "crossref"],
+    },
+    source_weights={
+        "semantic_scholar": 1.5,  # Prioritize academic sources
+        "arxiv": 1.5,
+        "web": 0.8,
+    },
+)
+```
+
+### Research Report Generation
+
+The research pipeline produces structured reports with proper citations:
+
+```python
+result = researcher.research(
+    "Compare reinforcement learning from human feedback (RLHF) "
+    "with direct preference optimization (DPO) for LLM alignment"
+)
+
+# The report includes:
+# - Executive summary
+# - Detailed findings organized by sub-topic
+# - Inline citations with source URLs
+# - Source bibliography
+# - Confidence assessment for each claim
+
+# Save the report
+result.save_markdown("rlhf_vs_dpo_report.md")
+result.save_html("rlhf_vs_dpo_report.html")
+```
+
+## Web Interface
+
+Local Deep Research includes a built-in web interface for interactive research sessions:
+
+```bash
+# Start the web UI
+local-deep-research --ui
+
+# Or specify host and port
+local-deep-research --ui --host 0.0.0.0 --port 5000
+```
+
+The web interface provides:
+
+- **Interactive research sessions**: Submit queries and watch the research process in real-time
+- **Source inspection**: Click through to original sources for each finding
+- **Research history**: Browse and re-examine previous research sessions
+- **Report export**: Download reports in markdown, HTML, or PDF format
+- **Configuration panel**: Adjust LLM and search settings without editing config files
+
+## Advanced Research Workflows
+
+### Iterative Research with Follow-Up Questions
+
+Build on previous research with follow-up queries:
+
+```python
+# Initial research
+result1 = researcher.research(
+    "Overview of graph neural networks for molecular property prediction"
+)
+
+# Follow-up that builds on context from the first query
+result2 = researcher.follow_up(
+    "Which of these approaches handle 3D molecular geometry?",
+    context=result1,
+)
+```
+
+### Batch Research
+
+Run multiple research queries in batch for systematic investigations:
+
+```python
+queries = [
+    "Attention mechanisms in protein structure prediction",
+    "Graph neural networks for drug-target interaction",
+    "Transfer learning approaches in computational chemistry",
+    "Benchmarks for molecular property prediction models",
+]
+
+results = researcher.batch_research(
+    queries,
+    parallel=True,
+    max_workers=4,
+)
+
+# Generate a comparative summary across all queries
+summary = researcher.synthesize(results)
+```
+
+### Fully Private Research Pipeline
+
+For maximum privacy, run everything locally with no external API calls:
+
+```bash
+# Use Ollama for LLM
+ollama pull llama3.1:70b
+
+# Use SearXNG for search (self-hosted)
+docker run -d --name searxng -p 8888:8080 searxng/searxng
+
+# Configure Local Deep Research
+export LDR_LLM_PROVIDER=ollama
+export LDR_LLM_MODEL=llama3.1:70b
+export SEARX_URL=http://localhost:8888
+export LDR_SEARCH_SOURCES=searxng,arxiv,pubmed,wikipedia
+
+# All queries now stay on your local machine
+local-deep-research "Your sensitive research query here"
+```
+
+## References
+
+- Repository: https://github.com/LearningCircuit/local-deep-research
+- Ollama: https://ollama.com/
+- SearXNG: https://github.com/searxng/searxng
+- Semantic Scholar API: https://api.semanticscholar.org/
+- arXiv API: https://info.arxiv.org/help/api/

package/skills/research/deep-research/open-researcher-guide/SKILL.md

@@ -0,0 +1,138 @@
+---
+name: open-researcher-guide
+description: "Open pipeline for generating deep research trajectories with LLMs"
+metadata:
+  openclaw:
+    emoji: "🔬"
+    category: "research"
+    subcategory: "deep-research"
+    keywords: ["OpenResearcher", "deep research", "research trajectory", "open pipeline", "literature synthesis", "LLM research"]
+    source: "https://github.com/GAIR-NLP/OpenResearcher"
+---
+
+# OpenResearcher Guide
+
+## Overview
+
+OpenResearcher is a fully open pipeline for long-horizon deep research trajectory synthesis. It breaks complex research questions into sub-questions, iteratively searches and reads literature, builds internal knowledge representations, and synthesizes comprehensive answers. Unlike single-shot approaches, it models the researcher's thought process — reading, questioning, connecting, and refining understanding over multiple rounds.
+
+## Pipeline Stages
+
+### 1. Question Decomposition
+
+```python
+from open_researcher import OpenResearcher
+
+researcher = OpenResearcher(llm_provider="anthropic")
+
+# Complex research question
+result = researcher.research(
+    "How do retrieval-augmented generation systems handle "
+    "knowledge conflicts between parametric and retrieved knowledge, "
+    "and what are the current mitigation strategies?"
+)
+
+# Automatically decomposes into sub-questions:
+# SQ1: What types of knowledge conflicts occur in RAG?
+# SQ2: How are conflicts detected?
+# SQ3: What resolution strategies exist?
+# SQ4: How effective are these strategies?
+```
+
+### 2. Iterative Search and Reading
+
+```python
+# Each sub-question triggers:
+# - Academic search (Semantic Scholar, arXiv)
+# - Paper reading (abstract + key sections)
+# - Evidence extraction
+# - Follow-up question generation
+
+# Configuration
+researcher = OpenResearcher(
+    search_backends=["semantic_scholar", "arxiv"],
+    max_iterations=5,           # Research rounds per sub-question
+    papers_per_iteration=10,    # Papers to read per round
+    follow_up_questions=True,   # Generate follow-up questions
+)
+```
+
+### 3. Knowledge Graph Building
+
+```python
+# Internally builds a knowledge representation:
+# - Claims linked to source papers
+# - Relationships between concepts
+# - Contradictions flagged
+
+# Access the knowledge graph
+kg = result.knowledge_graph
+print(f"Concepts: {len(kg.nodes)}")
+print(f"Relations: {len(kg.edges)}")
+print(f"Contradictions: {len(kg.contradictions)}")
+```
+
+### 4. Synthesis and Report
+
+```python
+# Multi-section synthesis
+report = result.report
+
+# Sections:
+# 1. Introduction and scope
+# 2. Sub-question answers with evidence
+# 3. Cross-cutting themes
+# 4. Open questions and future directions
+# 5. Full bibliography
+
+report.save("research_report.md")
+report.export_bibliography("refs.bib")
+```
+
+## Configuration
+
+```python
+researcher = OpenResearcher(
+    llm_provider="anthropic",
+    model="claude-sonnet-4-20250514",
+    search_config={
+        "backends": ["semantic_scholar", "arxiv"],
+        "max_results_per_query": 20,
+    },
+    reading_config={
+        "sections": ["abstract", "introduction", "methods", "conclusion"],
+        "max_tokens_per_paper": 3000,
+    },
+    synthesis_config={
+        "style": "academic",           # academic, technical, accessible
+        "include_contradictions": True,
+        "cite_inline": True,
+    },
+)
+```
+
+## Trajectory Inspection
+
+```python
+# Inspect the research trajectory
+trajectory = result.trajectory
+
+for step in trajectory:
+    print(f"Round {step.round}: {step.action}")
+    print(f"  Query: {step.query}")
+    print(f"  Papers read: {step.papers_read}")
+    print(f"  Key findings: {step.findings[:100]}...")
+    print(f"  Follow-ups: {step.follow_up_questions}")
+```
+
+## Use Cases
+
+1. **Literature surveys**: Comprehensive multi-round research
+2. **Research proposals**: Evidence gathering for grant applications
+3. **State-of-the-art reports**: Current landscape analysis
+4. **Tutorial generation**: Deep topic explanations with citations
+
+## References
+
+- [OpenResearcher GitHub](https://github.com/GAIR-NLP/OpenResearcher)
+- [GAIR-NLP Lab](https://github.com/GAIR-NLP)