@wentorai/research-plugins 1.0.0 → 1.2.0

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

@@ -0,0 +1,217 @@
+---
+name: tongyi-deep-research-guide
+description: "Open-source deep research agent by Alibaba for scholarly research"
+metadata:
+  openclaw:
+    emoji: "🔎"
+    category: "research"
+    subcategory: "deep-research"
+    keywords: ["deep-research", "alibaba", "tongyi", "agentic-rag", "scholarly-search", "open-source"]
+    source: "https://github.com/Alibaba-NLP/DeepResearch"
+---
+
+# Tongyi Deep Research Guide
+
+## Overview
+
+Tongyi DeepResearch is an open-source deep research agent developed by Alibaba's NLP team, with over 18,000 stars on GitHub. It implements an agentic research pipeline that iteratively searches, reads, reasons, and synthesizes information to produce comprehensive research reports. The system is designed to handle complex, multi-faceted research questions that require gathering evidence from multiple sources and reasoning across diverse information.
+
+Unlike simpler RAG (Retrieval-Augmented Generation) systems that perform a single search-and-answer cycle, DeepResearch uses an iterative approach where the agent dynamically decides what to search next based on what it has already found. This makes it particularly effective for research questions that require building up understanding incrementally, following citation chains, or exploring multiple angles of a topic.
+
+The project is notable for being one of the leading open-source alternatives to proprietary deep research tools. It supports multiple LLM backends, various search APIs, and can be customized for domain-specific research needs. For academic researchers, it offers a transparent and modifiable research pipeline where every step can be inspected, reproduced, and adapted.
+
+## Installation and Setup
+
+```bash
+# Clone the repository
+git clone https://github.com/Alibaba-NLP/DeepResearch.git
+cd DeepResearch
+
+# Install dependencies
+pip install -r requirements.txt
+
+# Or install with conda
+conda create -n deepresearch python=3.10
+conda activate deepresearch
+pip install -r requirements.txt
+```
+
+Configure your environment for the LLM and search backends:
+
+```bash
+# LLM configuration (supports multiple providers)
+export LLM_API_KEY=$LLM_API_KEY
+export LLM_BASE_URL=$LLM_BASE_URL
+export LLM_MODEL=qwen-max
+
+# Search API configuration
+export SEARCH_API_KEY=$SEARCH_API_KEY
+export SEARCH_ENGINE=bing  # or google, serper, tavily
+```
+
+For a fully local deployment with Ollama:
+
+```bash
+# Use local models
+export LLM_BASE_URL=http://localhost:11434/v1
+export LLM_MODEL=qwen2.5:72b
+export LLM_API_KEY=ollama
+```
+
+## Core Research Pipeline
+
+### The Iterative Research Loop
+
+DeepResearch follows a think-search-read-reflect loop that mimics how a human researcher works:
+
+1. **Think**: Analyze the research question and identify what information is needed
+2. **Search**: Formulate search queries and retrieve relevant documents
+3. **Read**: Extract and comprehend key information from retrieved documents
+4. **Reflect**: Evaluate whether enough information has been gathered or if further research is needed
+5. **Synthesize**: Compile findings into a structured, cited report
+
+```python
+from deep_research import DeepResearch
+
+# Initialize the research agent
+agent = DeepResearch(
+    llm_model="qwen-max",
+    search_engine="bing",
+    max_iterations=10,
+    max_sources=30,
+)
+
+# Run a research query
+result = agent.research(
+    query="What are the latest advances in multimodal large language models "
+          "and their applications in scientific research?",
+    output_format="markdown",
+)
+
+print(result.report)
+print(f"Sources consulted: {len(result.sources)}")
+print(f"Research iterations: {result.iterations}")
+```
+
+### Research Configuration
+
+Fine-tune the research behavior for different types of queries:
+
+```python
+config = {
+    "max_iterations": 15,          # Maximum research cycles
+    "max_sources_per_query": 10,   # Sources per search query
+    "min_relevance_score": 0.7,    # Minimum source relevance threshold
+    "enable_citation_tracking": True,  # Follow citation chains
+    "language": "en",              # Output language
+    "report_length": "detailed",   # brief, standard, or detailed
+}
+
+agent = DeepResearch(config=config)
+```
+
+### Supported Search Backends
+
+DeepResearch integrates with multiple search providers to cast a wide net:
+
+- **Bing Search API**: General web search with academic content
+- **Google Custom Search**: Configurable search with domain restrictions
+- **Tavily**: AI-optimized search API designed for research agents
+- **Serper**: Fast Google search results API
+- **SearXNG**: Self-hosted meta-search engine for privacy-focused deployments
+- **Semantic Scholar API**: Direct academic paper search (no API key required for basic access)
+
+```python
+# Configure multiple search backends for comprehensive coverage
+agent = DeepResearch(
+    search_engines=["bing", "semantic_scholar"],
+    search_strategy="parallel",  # Search all engines simultaneously
+)
+```
+
+## Advanced Features
+
+### Citation Chain Following
+
+DeepResearch can follow citation chains to discover related work:
+
+```python
+result = agent.research(
+    query="Foundational papers on attention mechanisms in neural networks",
+    enable_citation_tracking=True,
+    citation_depth=2,  # Follow citations up to 2 levels deep
+)
+```
+
+### Domain-Specific Research Profiles
+
+Create research profiles optimized for specific academic domains:
+
+```python
+# Biomedical research profile
+bio_config = {
+    "preferred_sources": ["pubmed", "biorxiv", "nature", "science"],
+    "search_engines": ["semantic_scholar", "bing"],
+    "terminology_mode": "technical",
+    "citation_format": "apa",
+}
+
+agent = DeepResearch(config=bio_config)
+result = agent.research(
+    "Recent developments in mRNA vaccine delivery mechanisms"
+)
+```
+
+### Streaming Progress
+
+Monitor the research process in real-time:
+
+```python
+async def stream_research():
+    agent = DeepResearch(llm_model="qwen-max")
+
+    async for event in agent.research_stream(
+        query="Quantum computing applications in drug discovery"
+    ):
+        if event.type == "thinking":
+            print(f"Thinking: {event.content}")
+        elif event.type == "searching":
+            print(f"Searching: {event.query}")
+        elif event.type == "reading":
+            print(f"Reading: {event.url}")
+        elif event.type == "report":
+            print(f"Final report:\n{event.content}")
+```
+
+## Research Workflow Integration
+
+### Combining with Academic Tools
+
+DeepResearch output can be integrated with standard academic tools:
+
+- Export reports as BibTeX-compatible references for LaTeX papers
+- Feed results into Zotero or Mendeley for reference management
+- Use the structured output as input for systematic review tools
+- Combine with local document collections for comprehensive literature coverage
+
+### Reproducibility
+
+Every research session can be fully reproduced:
+
+```python
+# Save the complete research trace
+result = agent.research(query="...", save_trace=True)
+result.save_trace("research_trace.json")
+
+# Replay a research session
+replayed = DeepResearch.replay("research_trace.json")
+```
+
+The trace includes all search queries, retrieved documents, LLM prompts and responses, and reasoning steps, enabling full transparency and reproducibility of the research process.
+
+## References
+
+- Repository: https://github.com/Alibaba-NLP/DeepResearch
+- Qwen model family: https://github.com/QwenLM/Qwen
+- Alibaba NLP group: https://github.com/Alibaba-NLP
+- Semantic Scholar API: https://api.semanticscholar.org/

package/skills/research/funding/eu-horizon-guide/SKILL.md

@@ -0,0 +1,244 @@
+---
+name: eu-horizon-guide
+description: "Navigate EU Horizon Europe funding programs and proposal writing"
+metadata:
+  openclaw:
+    emoji: "eu"
+    category: "research"
+    subcategory: "funding"
+    keywords: ["Horizon Europe", "EU funding", "ERC grants", "MSCA", "proposal writing", "European research", "consortium"]
+    source: "wentor-research-plugins"
+---
+
+# EU Horizon Europe Guide
+
+A skill for navigating the EU Horizon Europe Framework Programme, identifying relevant funding calls, understanding eligibility rules, building consortia, and writing competitive proposals. Covers the three-pillar structure, key instruments (ERC, MSCA, collaborative projects), budget planning, and the evaluation process.
+
+## Horizon Europe Structure
+
+### Three Pillars
+
+Horizon Europe (2021-2027) has a total budget of approximately EUR 95.5 billion, organized into three pillars plus a horizontal component.
+
+```
+Pillar I - Excellent Science (~EUR 25B):
+  - European Research Council (ERC): frontier research
+    - Starting Grants: 2-7 years post-PhD, up to EUR 1.5M
+    - Consolidator Grants: 7-12 years post-PhD, up to EUR 2M
+    - Advanced Grants: established leaders, up to EUR 2.5M
+    - Synergy Grants: 2-4 PIs, up to EUR 10M
+  - Marie Sklodowska-Curie Actions (MSCA): researcher mobility
+    - Postdoctoral Fellowships: individual mobility
+    - Doctoral Networks: structured PhD programs
+    - Staff Exchanges: inter-sector mobility
+  - Research Infrastructures: access to facilities
+
+Pillar II - Global Challenges and European Industrial Competitiveness (~EUR 53.5B):
+  - Six clusters:
+    1. Health
+    2. Culture, Creativity and Inclusive Society
+    3. Civil Security for Society
+    4. Digital, Industry and Space
+    5. Climate, Energy and Mobility
+    6. Food, Bioeconomy, Natural Resources, Agriculture, Environment
+  - Missions: Cancer, Climate Adaptation, Oceans, Cities, Soil
+
+Pillar III - Innovative Europe (~EUR 13.6B):
+  - European Innovation Council (EIC)
+  - European Innovation Ecosystems
+  - European Institute of Innovation and Technology (EIT)
+```
+
+### Finding Relevant Calls
+
+```
+Step-by-step call identification:
+
+1. Funding and Tenders Portal (primary source):
+   https://ec.europa.eu/info/funding-tenders/opportunities/portal/
+   - Search by keyword, cluster, instrument type
+   - Filter by open calls, upcoming deadlines
+   - Subscribe to email alerts for your topics
+
+2. Work Programme documents:
+   - Published every 2 years (2023-2024, 2025-2027)
+   - Detailed topic descriptions with expected outcomes
+   - Specify eligible participants, budget, project duration
+   - Read the FULL topic description, not just the title
+
+3. National Contact Points (NCPs):
+   - Each EU member state has NCPs for each cluster
+   - Free advice on eligibility, partner search, proposal review
+   - Often organize information days and matchmaking events
+
+4. Partner Search:
+   - CORDIS Partner Service
+   - Enterprise Europe Network
+   - NCP matchmaking events
+   - Conference networking sessions
+```
+
+## Proposal Writing
+
+### Structure of a Collaborative Proposal (RIA/IA)
+
+```
+Part A: Administrative forms (online)
+  - Participant information
+  - Budget tables
+  - Ethics issues table
+  - Call-specific declarations
+
+Part B: Technical description (PDF upload)
+  Section 1 - Excellence (scored):
+    1.1 Objectives and ambition
+    1.2 Methodology
+    1.3 Interdisciplinary/intersectoral approach (if relevant)
+
+  Section 2 - Impact (scored):
+    2.1 Project results and expected impacts
+    2.2 Communication, dissemination, exploitation
+    2.3 Summary of measures to maximize impact
+
+  Section 3 - Implementation (scored):
+    3.1 Work plan: work packages, tasks, deliverables, milestones
+    3.2 Management structure, milestones, critical risks
+    3.3 Consortium composition and partnership
+
+  Note: Page limits are STRICTLY enforced.
+  Typical limits: 40-50 pages for Part B.
+```
+
+### Writing Tips for High Scores
+
+```
+Excellence section:
+  - Start with the specific call topic text, show exact alignment
+  - State objectives as SMART goals (specific, measurable, achievable,
+    relevant, time-bound)
+  - Methodology: explain WHY this approach, not just WHAT you will do
+  - Include a Gantt chart showing work package timing
+  - Describe beyond-state-of-the-art contributions explicitly
+  - Use the exact terminology from the call topic
+
+Impact section:
+  - Use the Impact Canvas framework: outputs -> outcomes -> impacts
+  - Quantify impacts where possible (publications, patents, jobs,
+    policy changes, training beneficiaries)
+  - Name specific stakeholders who will use your results
+  - Include a credible exploitation plan (not just "we will publish")
+  - Show knowledge of the Target and Expected Impacts in the call
+
+Implementation section:
+  - Work packages should have clear, non-overlapping scope
+  - Each partner should lead at least one work package
+  - Include a dedicated management WP (WP1) and dissemination WP
+  - Risk register with likelihood, impact, and mitigation measures
+  - Pert chart or Gantt chart is expected
+```
+
+## Consortium Building
+
+### Partner Selection
+
+```
+Ideal consortium characteristics:
+
+Composition:
+  - 5-15 partners for a standard RIA (Research and Innovation Action)
+  - At least 3 partners from 3 different EU member states
+  - Mix of universities, research institutes, SMEs, industry, CSOs
+  - Include at least one partner from a widening country (bonus)
+
+Partner roles:
+  - Coordinator: strong project management track record
+  - Scientific leads: top researchers in the field
+  - Industry partners: exploitation and market access
+  - SMEs: agility, innovation, close to market
+  - End users: validation, real-world testing
+  - CSOs/NGOs: societal engagement, responsible innovation
+
+Red flags:
+  - Partners who do not respond to emails promptly
+  - Partners with no relevant publications or experience
+  - Duplicate competencies (no justification for overlap)
+  - Partners from the same country as the coordinator (too many)
+  - Partners who cannot provide co-funding if required
+```
+
+## Budget Planning
+
+### Eligible Costs
+
+```
+Budget categories in Horizon Europe:
+
+A. Personnel costs:
+   - Actual salary costs or unit costs (daily rates)
+   - Include employer social charges
+   - Personnel must be directly working on the project
+   - Time sheets or equivalent time recording required
+
+B. Subcontracting:
+   - Must be justified and identified in the proposal
+   - Cannot subcontract core scientific work
+   - Subject to best value-for-money procurement
+
+C. Purchase costs:
+   - Travel and subsistence
+   - Equipment (depreciation during project period)
+   - Other goods, works, and services
+
+D. Other cost categories:
+   - Financial support to third parties (cascading grants)
+   - Internally invoiced goods and services
+
+E. Indirect costs:
+   - Flat rate of 25% of eligible direct costs
+   - No need to justify or document
+
+Funding rates:
+  - RIA (Research and Innovation Action): 100% of eligible costs
+  - IA (Innovation Action): 70% (100% for non-profit)
+  - CSA (Coordination and Support Action): 100%
+  - MSCA: unit costs (fixed amounts per researcher-month)
+  - ERC: 100% + 25% indirect
+```
+
+## Evaluation Process
+
+### How Proposals Are Scored
+
+```
+Evaluation criteria and weights:
+
+Criterion         Weight  Threshold  Score Range
+-----------       ------  ---------  ----------
+Excellence        varies  3/5        0-5
+Impact            varies  3/5        0-5
+Implementation    varies  3/5        0-5
+Overall           -       10/15      sum of above
+
+Score descriptors:
+  5 = Excellent: successfully addresses all relevant aspects
+  4 = Very good: addresses the criterion very well but has
+      a small number of shortcomings
+  3 = Good: addresses the criterion well but with
+      a number of shortcomings
+  2 = Fair: broadly addresses the criterion but with
+      significant weaknesses
+  1 = Poor: fails to convincingly address the criterion
+  0 = Not applicable or missing
+
+Process:
+  1. Individual remote evaluation (4-5 evaluators per proposal)
+  2. Consensus meeting (evaluators discuss and agree scores)
+  3. Panel review (ranking, budget allocation)
+  4. Ethics review (if flagged)
+  5. Grant Agreement preparation (6-9 months after deadline)
+
+Typical success rate: 10-15% for most calls
+ERC success rate: 10-12% for Starting Grants
+```
+
+Horizon Europe proposals require significant investment of time and resources. Starting preparation 6 to 12 months before the deadline, engaging National Contact Points early, and assembling a balanced consortium with complementary expertise are the strongest predictors of success, alongside the scientific quality of the proposed research itself.