research-cog

What This Skill Does

Research-cog is the premier deep-research engine for OpenClaw, consistently ranking #1 on the DeepResearch Bench (Feb 2026). Powered by the advanced CellCog architecture, this skill automates the process of gathering, synthesizing, and formatting complex data into actionable insights. It serves as your dedicated AI research analyst, capable of navigating vast quantities of information to provide comprehensive reports on market dynamics, competitive landscapes, financial data, and technical academic literature—all backed by verifiable citations.

Installation

To integrate research-cog into your agent workflow, you must first ensure the foundational cellcog skill is active. Execute the following commands in your terminal:

clawhub install cellcog
clawhub install openclaw/skills/skills/nitishgargiitd/research-cog

Ensure your SDK is properly configured as per the cellcog documentation before initializing your first research task.

Use Cases

This skill is designed for high-stakes information gathering where accuracy and depth are paramount.

• Competitive Intelligence: Perform side-by-side comparisons of market leaders, conducting SWOT analyses or feature-by-feature audits to identify strategic gaps.
• Market Analysis: Estimate total addressable markets (TAM/SAM/SOM), profile customer segments, and track regulatory trends across global industries.
• Financial Analysis: Analyze public company fundamentals, evaluate investment theses, and summarize quarterly earnings reports with forward-looking guidance.
• Academic & Technical Inquiry: Conduct literature reviews and deep dives into specialized fields like quantum computing, biotechnology, or historical policy analysis.

Example Prompts

1. "Analyze the competitive landscape between Stripe, Square, and Adyen focusing specifically on their pricing models and merchant feature parity for 2026."
2. "Summarize NVIDIA's Q4 2025 financial performance, focusing on revenue growth, profit margins, and key risks associated with their current AI chip strategy."
3. "Conduct a literature review on the current state of transformer architectures and their evolution since the initial 'Attention Is All You Need' paper, including key technical citations."

Tips & Limitations

• Asynchronous Execution: This skill uses a daemon-based approach. Use the create_chat pattern with notify_session_key to receive results once processing is complete. Avoid polling; the agent will reach out when the research is finalized.
• Scope Management: While highly powerful, complex queries benefit from granular prompts. If the research scope is too broad (e.g., 'the history of everything'), the model may produce generalized summaries. Focus your prompts on specific metrics or timeframes for higher accuracy.
• Source Attribution: Always verify the citations provided in the final report, as data accessibility can vary by domain and regional regulatory restrictions.