A sophisticated multi-source research methodology for AI agents that implements a 4-tier credibility system and automated evidence tracking.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install verified-research
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 verified-research 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 Skill?
The Deep Research Skill is a high-integrity methodology designed for Openclaw Skills users who require more than just surface-level answers. It transforms an AI agent into a digital researcher capable of executing multi-dimensional search plans, verifying facts across at least three independent sources, and maintaining a structured evidence cache.
This skill is particularly powerful for technical, academic, or market research where accuracy and source attribution are non-negotiable. By utilizing a 3-day intelligent cache and a 4-tier credibility ranking (from T1 official docs to T4 social media), it ensures that every conclusion is backed by verifiable data rather than model hallucinations.
Deep Research Skill Use Cases
- Investigating emerging technology trends or open-source project health.
- Performing competitive analysis with multi-source cross-verification.
- Fact-checking complex claims using T1 academic or official documentation.
- Executing multi-day research projects that require persistent state and context.
- Automating the generation of structured research reports with full source attribution.
How Deep Research Skill Works
- Intent Recognition: The skill triggers when it detects research or investigation keywords.
- Decomposition: The agent breaks the main topic into sub-questions and defines research boundaries.
- Source Tiering: It prioritizes T1 (Official) and T2 (Authority) sources over T3/T4 community content.
- Evidence Collection: Every finding is recorded as a claim card in a local cache directory.
- Cross-Verification: The agent performs mandatory re-searching with different keywords to verify consistency.
- Report Generation: A final Markdown report is synthesized using the evidence cards and manifest.
- Persistent Cache: Data is stored for 3 days, allowing users to continue research sessions seamlessly.
- Archival: Inactive research is summarized into MEMORY.md before the cache is cleared.
Deep Research Skill Setup
To initialize the research environment, navigate to the scripts directory within your Openclaw Skills workspace and execute the research runner:
cd /root/.openclaw/workspace/skills/deep-research/scripts
TOPIC="your_research_topic" SESSION_ID="optional_id" bash research.sh
For ongoing maintenance and cleanup of the 3-day cache, you can run:
bash cleanup.sh
Deep Research Skill Data Schema & Taxonomy
The skill organizes its workflow within a structured cache system to ensure transparency and reproducibility:
| File/Folder | Description |
|---|---|
/claims/ |
Individual evidence cards (claim_NNN.md) containing conclusions and sources. |
/rounds/ |
Logs for each research iteration (round_NNN.md). |
manifest.json |
Index of all evidence and credibility statistics. |
report_final.md |
The comprehensive synthesized research report. |
MEMORY.md |
Long-term storage for research summaries and core conclusions. |
Deep Research Skill Advanced Features
- Intelligent 3-Day Cache: Prevents redundant API calls while allowing for session continuation.
- Automated Source Tiering: Assigns weights to information based on its origin (T1 to T4).
- Quality Control Gates: Automatically blocks report generation if cross-verification requirements are not met.
- Self-Improving Integration: Logs research quality and lessons learned to .learnings/research-log.md to refine future search strategies.
- Conflict Resolution: Explicitly highlights and flags contradictory information from different sources.
SKILL.md
Loading
Related Openclaw Skills
Polymarket BTC 5m Trading Bot
A high-frequency AI trading bot designed to automate BTC prediction market trades on Polymarket using dual-signal logic and real-time decision-making.
Documentation Organization
A systematic framework for restructuring project documentation to optimize AI context usage and prevent information drift.
Kalshi Fed Futures Arb Trader
An automated trading skill that captures alpha by identifying pricing gaps between institutional CME FedWatch futures and Kalshi prediction markets.
Kalshi Fed Data Reaction Trader
An automated trading agent that executes Kalshi Fed rate market positions by analyzing market-implied CPI signals and macro data sensitivity.
Patchright Browser Automation
A powerful browser automation tool based on Patchright that enables AI agents to interact with local web applications while bypassing bot detection.
pytest-test-master
A specialized technical guide and CLI tool designed to solve complex pytest challenges through expert fixture management, mocking patterns, and CI/CD coverage integration.
Featured*
