A high-fidelity research engine for AI agents that prioritizes exhaustive investigation and academic-style reporting over superficial speed.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install deep-research-2
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install deep-research-2 using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is Deep Research?
Deep Research is a specialized capability designed for Openclaw Skills that transforms a standard AI model into a rigorous academic assistant. Unlike basic search tools that provide surface-level summaries, this skill emphasizes a strict hierarchy of information, starting with authoritative APIs and comprehensive web fetching. It is specifically engineered to handle complex, multi-faceted queries by decomposing them into conceptual maps, ensuring that every claim is backed by verified citations and rigorous cross-referencing.
By integrating this tool into your workflow, you leverage a systematic approach to knowledge acquisition where completeness and accuracy are the primary metrics of success. The output is a formal markdown report that mirrors the depth of human-led scholarly investigation. This approach within the ecosystem of Openclaw Skills ensures that the resulting documents are objective, evidence-based, and suitable for professional or academic environments.
Deep Research Use Cases
- Generating comprehensive academic-style reports on emerging technologies
- Conducting multi-lingual market research to identify global trends and data points
- Performing deep-dive competitive analysis for high-stakes business intelligence
- Investigating complex regulatory frameworks across different international jurisdictions
- Creating evidence-based whitepapers with fully verified citations and reference lists
How Deep Research Works
- Topic Decomposition: The agent identifies core research questions and breaks them down into sub-topics and key concepts.
- Knowledge Grounding: It identifies and defines unknown terms through authoritative sources to establish a solid conceptual foundation.
- Staged Searching: Utilizing Openclaw Skills, the agent performs progressive web searches, fetching full page content rather than relying on snippets.
- Multi-Lingual Verification: Information is cross-checked in at least two languages to eliminate linguistic bias and capture region-specific data.
- Sectional Drafting: Each chapter of the report is drafted as an individual markdown file to maintain high detail and prevent context loss.
- Final Assembly: The agent compiles all drafts into a cohesive, formal report with a complete, alphabetized reference section.
Deep Research Setup
To initialize this skill within your environment, ensure you have the necessary research templates and methodology files in your assets and references directories.
# Move the deep-research directory to your skills path
cp -r ./skills/deep-research ~/.openclaw/skills/
# Configure your search and fetch providers
openclaw config set search_api_key YOUR_API_KEY
Once configured, the skill can be triggered using phrases like "comprehensive investigation" or "academic research" within any interface supporting Openclaw Skills.
Deep Research Data Schema & Taxonomy
The Deep Research skill manages information through a structured taxonomy of markdown files to ensure data integrity during the long-form generation process.
| File Type | Purpose | Location |
|---|---|---|
| Section Drafts | Individual markdown files for each report section (e.g., draft_intro.md) |
/drafts/ |
| Final Report | The complete, assembled multi-thousand word academic document | /output/ |
| Methodology | Detailed guidelines for search logic and critical thinking | references/ |
| Citation Guide | Instructions for maintaining consistent reference formats | references/ |
| Report Template | Structural boilerplate for the final output document | assets/ |
Deep Research Advanced Features
- Sequential Entity Processing: Investigates multiple subjects individually and completely before moving to the next subject to ensure deep coverage
- Multi-Lingual Search Strategy: Automatically translates queries to overcome information siloing in non-English data sets
- Information Priority Hierarchy: Enforces a strict protocol that favors structured Data Source APIs over general web search
- Automated Reference Management: Generates formal, clickable citation lists for all substantive claims and statistics
- Context-Preservation Drafting: Utilizes separate file storage for sections to ensure the final report maintains maximum word count and detail
SKILL.md
Loading
Related Openclaw Skills
Feishu Forwarded Message Reader
A specialized tool for extracting and displaying the full content of merged and forwarded messages within the Feishu platform.
AI Agent Team Manager
A professional orchestration system for coordinating multiple AI agents using the Otter Camp methodology within the Openclaw Skills ecosystem.
x402 Private Web Tools
A privacy-first suite of web tools allowing AI agents to search, scrape, and capture the web using USDC micropayments on the Base network.
x402 Client
Automate USDC micro-payments for HTTP API requests on the Base Sepolia network.
MediaWiki Wikitext
Generate and edit high-quality content using MediaWiki wikitext markup for Wikipedia and other wiki-based platforms.
Prompt Engineering
A comprehensive framework for transforming vague requests into precise, high-performance LLM prompts through systematic analysis and iterative refinement.
Featured*
