A framework for breaking down complex tasks into verifiable atomic steps using first principles thinking.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install principles-agent
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 principles-agent 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 Principles Agent?
Principles Agent is a sophisticated reasoning and decomposition framework built for the Openclaw Skills ecosystem. Inspired by the miltonian/principles design philosophy, it empowers users to solve complex problems by stripping away assumptions and returning to foundational axioms. Unlike standard agents that attempt to solve problems in a single pass, this skill implements an iterative architecture that separates the logic of task planning from the actual execution of tasks.
By leveraging first principles thinking, the skill ensures that every sub-task is grounded in verifiable facts and logical necessity. It functions as an architectural layer for your AI workflows, providing a structured approach to project design, system engineering, and deep analytical research. Within the broader context of Openclaw Skills, it acts as a high-level orchestrator that maintains logic integrity while delegating I/O operations to the main agent for maximum security and control.
Principles Agent Use Cases
- Designing complex software architectures or large-scale engineering projects from the ground up.
- Conducting deep-dive research on multifaceted technical or philosophical problems to find root causes.
- Orchestrating multi-step AI agent collaborations where step-by-step verification is critical for success.
- Refining vague project requirements into concrete, actionable, and atomic task lists.
How Principles Agent Works
- Requirement Clarification: Receives the user goal and extracts success criteria and constraints.
- First Principles Deconstruction: Derives foundation facts and axioms by stripping away experience-based assumptions.
- Atomic Task Decomposition: Breaks down the derived truths into the smallest possible verifiable sub-tasks.
- Iterative Refinement: Automatically checks sub-tasks for feasibility and alignment, optimizing them through feedback loops.
- Dependency Sorting: Utilizes topological sorting to determine the logical order of execution and identify parallel tasks.
- Execution Dispatch: Routes sub-tasks to the main Openclaw Agent for LLM processing and validates each result.
- Global Integration: Merges all sub-task outputs into a cohesive final delivery.
- Final Acceptance: Validates the integrated result against the original goal to ensure complete project fulfillment.
Principles Agent Setup
To begin using this skill within your Openclaw Skills environment, ensure you have the repository structure correctly mapped. Since the skill follows a minimal permission model, no external API keys or network configurations are required directly by the skill itself.
# Navigate to your Openclaw directory and add the skill
# Use the command line interface to trigger the principles logic
principles "Your complex goal or system design description"
# To save the structured output to a markdown file
principles "Your project goal" --output project_plan.md
# To limit the depth of the iterative refinement process
principles "Complex problem" --max-iterations 4
Principles Agent Data Schema & Taxonomy
The Principles Agent organizes its internal logic through a structured set of Python modules and outputs refined Markdown reports. It maintains strict data types for task management and truth derivation.
| Module | Functionality |
|---|---|
| truth_deriver.py | Manages the derivation of foundational facts and axioms. |
| task_breaker.py | Handles the tree-based decomposition of goals into atomic units. |
| dependency_sorter.py | Manages the DAG (Directed Acyclic Graph) for task execution order. |
| validator.py | Implements layered quality checks for both task structure and execution results. |
| integrator.py | Compiles individual sub-task results into the final output format. |
Principles Agent Advanced Features
- Topological dependency management to prevent circular reasoning and enable efficient task sequencing.
- Secure isolation architecture where the skill never accesses the network or environment variables directly.
- Iterative feedback loops that allow the agent to self-correct during the decomposition phase.
- Seamless integration with the main Openclaw Skills executor for handling external LLM calls and complex I/O.
SKILL.md
Loading
Related Openclaw Skills
Goal Manager v2.0
A comprehensive goal management assistant supporting OKR and SMART frameworks for tracking professional and personal progress.
File Super Assistant
A professional toolset for programmatic document generation and transforming robotic AI text into natural, human-like writing.
Hotel Comparison Assistant (fb-hotel-comparison-skill)
A real-time hotel price aggregator and comparison engine designed to find the best travel deals across major OTA platforms.
Shortplay Generator
An automated workflow for generating viral short-form drama scripts complete with optimized AI video production prompts.
Movie Recommender AI Assistant
A sophisticated AI-driven movie discovery and viewing management tool that leverages Douban data for hyper-personalized recommendations.
Problem Mapper
A sophisticated problem-diagnosis framework that uses system thinking and coaching-style guidance to transform complex ambiguity into clear action plans.
Featured*
