A sophisticated AI orchestration framework that automates task planning, execution, and quality review based on request complexity.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-system
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 agent-system 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 OpenClaw Agent System?
The OpenClaw Agent System is a core intelligent scheduling framework designed for Openclaw Skills. It acts as the brain of an AI agent, dynamically routing tasks through a pipeline of specialized components including an Orchestrator, Planner, Executor, and Reviewer. By evaluating the complexity of user requests, the system ensures that simple tasks are handled efficiently while complex, multi-step requirements undergo rigorous planning and quality assurance to maintain high-performance standards across Openclaw Skills.
OpenClaw Agent System Use Cases
- Deep analytical tasks requiring multi-step reasoning and root cause identification.
- Automated code generation and system design involving complex architecture within Openclaw Skills.
- Quality-critical content generation where a Reviewer and Self-Heal cycle are necessary.
- Strategic planning and scheduling tasks that require breaking down high-level goals into actionable steps.
How OpenClaw Agent System Works
- The Orchestrator receives user input and identifies the intent, such as analysis, coding, or planning.
- A complexity assessment is performed; tasks are rated on a scale of 1 to 10 based on specific technical indicators found in Openclaw Skills.
- The system routes the task based on the score: simple tasks go directly to the Executor, while mid-range tasks include a Reviewer stage.
- For complex tasks with a score higher than 5, the Planner decomposes the request into multiple sub-tasks.
- The Reviewer scores the output; if the quality score is below 70, the Self-Heal mechanism triggers a retry with refined reasoning.
- Final results are delivered in a structured JSON format including execution metrics and quality scores to the Openclaw Skills environment.
OpenClaw Agent System Setup
Install the core orchestrator components into your project environment to begin using Openclaw Skills:
# Integrate the orchestrator into your Openclaw Skills workspace
cp -r src/orchestrator.js ./your-project/src/
Configure your environment variables to allow for multi-step LLM calls as required by the Planner and Reviewer modules.
OpenClaw Agent System Data Schema & Taxonomy
The system communicates via a standardized JSON schema to maintain consistency across Openclaw Skills.
| Key | Type | Description |
|---|---|---|
| success | Boolean | Indicates if the execution was successful. |
| content | String | The final output or result of the task. |
| plan | Object | Contains intent, complexity, and the list of decomposed tasks. |
| metrics | Object | Tracks token usage, quality scores, and self-heal triggers. |
OpenClaw Agent System Advanced Features
- Automated Self-Healing: Intelligent retry strategies including detailed reasoning and reverse reasoning for failed tasks within Openclaw Skills.
- Degraded Mode: Ensures system reliability by providing simplified results when multiple retry attempts fail.
- Structured JSON Logging: Comprehensive audit trails for every module including timestamps and event levels.
- Dynamic Complexity Routing: Optimizes resource usage by bypassing heavy planning for simple, low-complexity queries in Openclaw Skills.
SKILL.md
Loading
Related Openclaw Skills
AI Redaction
A natural language-driven document redaction tool designed to securely mask sensitive information in files like PDFs and images.
Audit Verification Pipeline
A deterministic three-tier validation workflow that converts probabilistic AI security findings into verified, high-quality audit deliverables using Foundry.
OpenCLI
OpenCLI is a universal command-line hub that transforms websites, Electron applications, and local tools into structured, deterministic APIs for users and AI agents.
AI Gaming Content and Monetization Machine
An end-to-end AI agent skill that researches gaming trends, writes viral scripts, and produces complete videos for YouTube and TikTok monetization.
Nexus Talent Assessor
An AI-powered HR agent designed for competency-based evaluations, skill gap analysis, and automated career development planning.
Nexus Market Researcher
A strategic market research agent designed to perform deep competitive analysis and market sizing across multiple industry verticals.
Featured*
