A high-performance reasoning engine that automatically selects the best logic strategy for AI agents with parallel execution and intelligent caching.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install unified-reasoning
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 unified-reasoning 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 Unified Reasoning Engine?
The Unified Reasoning Engine is a sophisticated logic orchestration layer designed for the Openclaw Skills ecosystem. It serves as a single entry point for complex problem-solving, automatically selecting and applying the most effective reasoning strategy—such as Chain of Thought (CoT), Tree of Thoughts (ToT), or Graph of Thoughts (GoT)—based on the specific requirements of the task.
By leveraging the Framework of Thoughts (FoT) optimization, this skill introduces parallel execution of thought branches and intelligent intermediate result caching. This allows developers to achieve 2-5x speedups on complex reasoning tasks compared to sequential processing. Whether you are building autonomous agents or complex decision-making systems, this engine ensures high-accuracy outputs with optimized compute usage for all your Openclaw Skills.
Unified Reasoning Engine Use Cases
- Automating complex task prioritization based on dynamic environmental variables.
- Verifying the accuracy of mathematical or logical outputs using Self-Consistency voting.
- Synthesizing divergent ideas into a single coherent strategy via Graph of Thoughts.
- Powering AGI-level decision-making processes for autonomous agent controllers.
- Reducing latency in multi-step reasoning through parallel branch execution and result caching.
How Unified Reasoning Engine Works
- The engine receives a problem via the
Invoke-Reasoningcommand and passes it to the Meta-Reasoning Layer. - The system analyzes problem characteristics like complexity, domain keywords, and the need for synthesis or verification.
- Based on this analysis, the engine selects the optimal strategy (e.g., Meta-Reasoning, Tree of Thoughts, or Hybrid).
- The Framework of Thoughts (FoT) optimizer initiates parallel execution of reasoning branches or nodes.
- The engine checks the global reasoning cache to reuse previously computed intermediate thoughts.
- Results from various branches are synthesized or voted upon to determine the final solution.
- A structured JSON-compatible object is returned, including a confidence score and execution metadata.
Unified Reasoning Engine Setup
To integrate this engine into your environment for Openclaw Skills, follow these steps:
# Navigate to your local skills directory
cd path/to/openclaw/skills
# Clone or download the unified-reasoning repository
git clone https://github.com/openclaw/unified-reasoning.git
# Source the reasoning engine in your PowerShell or CLI environment
. ./unified-reasoning/reasoning-engine.ps1
# (Optional) Customize the configuration in the YAML file
nano ./unified-reasoning/config.yaml
Unified Reasoning Engine Data Schema & Taxonomy
The Unified Reasoning Engine returns a comprehensive data object that details the logic lifecycle for Openclaw Skills.
| Attribute | Type | Description |
|---|---|---|
strategy |
String | The reasoning methodology employed (e.g., GoT, ToT). |
solution |
String | The final output or the most optimal path found. |
confidence |
Float | A confidence score from 0.0 to 1.0. |
duration |
Integer | Total processing time in milliseconds. |
metThreshold |
Boolean | Indicates if the confidence score met the pre-defined threshold. |
metaReasoning |
Object | Specific data regarding why a particular strategy was chosen. |
Unified Reasoning Engine Advanced Features
- Framework of Thoughts (FoT) for massive speedups via multi-threaded branch generation.
- Intelligent logic caching to avoid redundant LLM calls for recurring reasoning patterns.
- Adaptive strategy selection that switches between CoT, ToT, and GoT based on word count and intent.
- Hybrid reasoning modes that combine multiple strategies (e.g., ToT + Self-Consistency) for maximum accuracy.
- Direct integration hooks for AGI controllers to automate high-level decision making.
SKILL.md
Loading
Related Openclaw Skills
Tree of Thoughts Reasoning
An advanced reasoning skill for Openclaw Skills that explores multiple solution branches, evaluates them based on custom criteria, and selects the most optimal path.
Task Orchestra
A powerful orchestration engine for managing multi-agent workflows, complex task dependencies, and parallel execution logic.
SkillOrchestra
SkillOrchestra is a high-performance routing framework that intelligently directs tasks to AI agents based on explicit skill competence and cost-efficiency metrics.
Graph of Thoughts (GoT) Reasoning
An advanced multi-path reasoning framework that enables AI agents to explore, combine, and synthesize complex solutions using non-linear graph structures.
World Model for AGI
A production-grade environment understanding and causal reasoning engine designed for high-performance AGI simulations and decision support.
OPC (One-Person Company) Framework
A comprehensive 14-skill AI agent framework designed to help solo entrepreneurs manage the entire business lifecycle from ideation to operations.
Featured*
