An iterative meta-skill that coordinates multiple search engines and academic databases to produce scientific-grade research reports.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install deep-researcher
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-researcher 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 Researcher?
Deep Researcher is a sophisticated meta-skill designed for developers and researchers who require more than a single-pass search. It functions as a process controller for Openclaw Skills, orchestrating a complex workflow of hypothesis testing, source-quality assessment, and contradiction resolution. By integrating multiple specialized tools, it ensures that AI-generated reports are grounded in both current web data and peer-reviewed academic literature.
This skill is particularly valuable for long-range forecasting, policy analysis, and technical deep-dives where accuracy and source traceability are non-negotiable. Using Openclaw Skills like this allows for a modular yet highly structured approach to information retrieval, transforming raw search results into actionable, footnoted intelligence formatted as scientific Markdown.
Deep Researcher Use Cases
- Forecasting market trends and labor impacts for specific time horizons like 2030.
- Conducting academic-grade literature reviews using Semantic Scholar mappings via literature-search.
- Resolving conflicting data points between institutional reports and news cycles using multi-agent arbitration.
- Generating scientific-style Markdown reports with full citation metadata and footnotes.
- Decomposing broad, complex questions into testable sub-questions for systematic analysis.
How Deep Researcher Works
- Decompose the research topic into specific sub-questions and measurable hypotheses to establish a clear roadmap.
- Perform a broad landscape scan using Tavily to identify major claims, key institutions, and relevant long-form content.
- Conduct an academic evidence pass to validate or refine web-based claims against scholarly papers and peer-reviewed sources.
- Trigger contradiction resolution to arbitrate conflicting data points based on methodology strength and source recency.
- Synthesize findings into a structured Markdown report, applying strict quality gates to ensure every major claim is supported by multiple independent sources.
Deep Researcher Setup
Ensure you have Node.js, curl, and jq installed. Then, install the required Openclaw Skills components using the following commands:
npx -y clawhub@latest install deepresearchwork tavily-search literature-search perplexity-deep-search
Configure your environment variables to allow the skill to access search APIs:
export TAVILY_API_KEY="your_key_here"
export PERPLEXITY_API_KEY="your_key_here"
Verify the installation and check script help files:
npx -y clawhub@latest list
node skills/tavily-search/scripts/search.mjs --help
Deep Researcher Data Schema & Taxonomy
The skill organizes its research output into a strict scientific Markdown structure. It manages data according to the following taxonomy:
| Component | Description |
|---|---|
| Footnotes | Full citation metadata including Author, Title, Venue, DOI, and URL |
| Confidence Level | High, Medium, or Low assessment provided for every major claim |
| Source Threshold | Minimum of 2 independent sources required to validate a claim |
| Methodology | Explicit disclosure of search parameters, date ranges, and tool mapping |
| Contradictions | A dedicated section for unresolved conflicts and arbitration logic |
Deep Researcher Advanced Features
- Multi-round iterative loops that allow for deep, recursive evidence gathering beyond a single query.
- Explicit contradiction arbitration rules that prioritize method-transparent sources over opaque data.
- Semantic Scholar mapping for deep academic integration, ensuring scholarly rigor in every report.
- Specialized search modes including reasoning and research modes for targeted fact-checking.
- Automated quality gates that prevent the generation of reports if evidence thresholds or source diversity requirements are not met.
SKILL.md
Loading
Related Openclaw Skills
Human Masked Content Creator
A sophisticated meta-skill that orchestrates multiple specialized agents to produce authentic, high-signal content that avoids robotic AI patterns.
Cold Outreach Hunter
An advanced orchestration meta-skill that automates the end-to-end B2B cold outreach pipeline from lead sourcing to personalized sequence generation.
Autonomous Bookkeeper
A sophisticated meta-skill that automates the pre-accounting lifecycle by orchestrating email intake, OCR data extraction, payment verification, and accounting record creation.
Qwen3-TTS MLX
A high-performance local text-to-speech engine optimized for Apple Silicon that supports voice cloning and natural language voice design.
DeFi Sniper
A high-speed orchestration skill for early token detection, risk assessment, and automated execution on Solana and Base.
FinTS Banking Agent
A secure interface for AI agents to interact with German bank accounts via the FinTS standard using the fints-agent-cli.
METADATA
Featured*
