Hyperthink is a high-performance research engine that utilizes a triple-perspective pipeline to generate synthesized, 80,000-word deep-dives with embedded confidence tagging.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install hyperthink
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 hyperthink 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 Hyperthink?
Hyperthink is a sophisticated research framework designed for Openclaw Skills that moves beyond simple LLM queries. It employs a 6-stage autonomous pipeline to analyze any topic through three distinct lenses: the Optimist, the Analyst, and the Critic. By running dozens of parallel deep-dives, it ensures that final reports are not just comprehensive, but fact-checked and balanced against bias.
This skill is engineered for complex decision-making environments where technical truth and risk assessment are paramount. It transforms a single user trigger into a massive, unattended workflow that produces executive briefs and comprehensive synthesis documents, complete with confidence tiers and net assessments. While it operates over an 8-18 hour window to optimize for cost via batch processing, the result is a professional-grade research artifact equivalent to weeks of human effort.
Hyperthink Use Cases
- Evaluating investment theses or market entry strategies with multi-agent adversarial analysis.
- Building foundational research for new technical projects or product developments.
- Conducting deep-dives into regulatory or legal landscapes where risk identification is critical.
- Generating high-fidelity, confidence-tagged whitepapers or internal knowledge bases.
- Comparing complex architectural options in software engineering or business operations.
How Hyperthink Works
- The user triggers the /hyperthink command and undergoes an interrogation flow to define scope, goals, and constraints.
- Stage 1 uses Claude Opus to generate a Master Brief and 36 persona-specific section prompts.
- Stage 2 executes 36 parallel deep-dives using the Optimist, Analyst, and Critic personas via the Anthropic Batch API.
- Stage 3a performs a Trifecta Audit where the Analyst reviews and confidence-tags claims from all perspectives.
- Stage 3b synthesizes the audited data into a massive, unified narrative (up to 80,000 words).
- Stage 4 produces a high-level executive brief in both Markdown and DOCX formats for final delivery.
Hyperthink Setup
To integrate Hyperthink into your Openclaw Skills environment, ensure the interrogate skill is installed and configure your environment variables.
# Required Environment Variables
export ANTHROPIC_API_KEY="your_key_here"
# Optional Notification Settings
export TELEGRAM_BOT_TOKEN="your_token"
export TELEGRAM_CHAT_ID="your_id"
# Prepare Directory Structure
mkdir -p /data/hyperthink/
# Install Dependencies for DOCX generation
pip install python-docx
Hyperthink Data Schema & Taxonomy
Hyperthink organizes data within a dedicated slug-based directory to maintain state across the long-running pipeline:
| File/Folder | Description |
|---|---|
pipeline-state.json |
The source of truth for stage tracking and Job IDs. |
sections/ |
Contains 36 individual deep-dive Markdown files categorized by persona. |
audit/ |
Stores 12 trifecta comparison documents with confidence tagging. |
synthesis.md |
The massive unified narrative (50k-80k words). |
brief.docx |
The final formatted executive deliverable. |
Hyperthink Advanced Features
- Triple-Persona Bias Mitigation: Explicitly forces the AI to fight confirmation bias through adversarial Critic and Optimist agents.
- Batch API Cost Optimization: Utilizes Anthropic's batch processing to reduce costs by 50% for large-scale runs.
- Idempotent Cron Polling: A mandatory 9-step sequence ensures the pipeline survives restarts and only proceeds when batch jobs are confirmed.
- Automated DOCX Formatting: Includes specialized scripts to convert complex Markdown, including emoji-based confidence tags, into professional documents.
- Confidence Tiers: Employs a visual tagging system (High/Medium/Disputed) throughout the synthesis to highlight evidentiary strength.
SKILL.md
Loading
Related Openclaw Skills
Self-Improving + Proactive Decision Making Agent
A proactive decision-support agent that learns your personal risk profile and framework preferences to provide structured, self-improving analysis.
Generation Text Generator
An AI-powered video generation tool that converts text prompts and documents into fully rendered 1080p MP4 videos without manual editing.
Hallucination Detector
An original AI safety tool designed to identify and correct hallucinations in AI-generated code, ensuring technical accuracy and reliability.
Zyka AI Media Generation
A unified command-line interface for generating professional-grade AI videos, images, and speech using over 40 industry-leading models.
Free Jesus AI
A powerful AI video generator for creating faith-based content and scripture messages using simple text prompts or media uploads.
Pilot Service Agents: Vehicles
An automotive data skill providing direct access to NHTSA vehicle records including VIN decoding, recall history, and consumer complaints.
Featured*
