A comprehensive runtime system for AI agents that manages tool lifecycles, permissions, and context optimization.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install agent-runtime
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-runtime 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 Agent Runtime?
The Agent Runtime is a sophisticated system designed to orchestrate the lifecycle of intelligent agents. It provides a robust framework for integrating tool registries, enforcing permission policies, and intercepting tool execution via hooks. By leveraging this tool within the ecosystem of Openclaw Skills, developers can build specialized sub-agents tailored for tasks like code exploration, strategic planning, or automated verification, all while maintaining strict control over token usage and context compression.
Agent Runtime Use Cases
- Orchestrating specialized sub-agents for complex coding tasks.
- Enforcing security policies and tool permissions during agent execution.
- Monitoring and tracking token usage across multiple agent sessions.
- Implementing pre- and post-execution hooks for debugging and logging.
- Compressing long conversation histories to optimize context window efficiency.
How Agent Runtime Works
- The user message is received and processed by the core Agent Runtime.
- The system queries the ToolRegistry to locate and route required functions.
- PermissionPolicy checks are performed to ensure the agent has authority to execute specific tools.
- The Hook Runner executes PreToolUse logic before the primary action occurs.
- The Executor runs the tool and passes the output to the PostTool hook for final processing.
- The SessionCompactor optimizes the context while UsageTracker records token consumption before returning the result.
Agent Runtime Setup
To integrate this into your environment, ensure the core runtime script is accessible within your project structure. Using Openclaw Skills allows for seamless script management.
# Navigate to your project directory
cd your-project-path
# Ensure the runtime script is located in the scripts folder
ls ./scripts/agent-runtime.mjs
Initialize the runtime in your JavaScript environment:
import { AgentRuntime } from './scripts/agent-runtime.mjs';
const agent = new AgentRuntime({
agentType: 'explore',
maxTokens: 100000
});
Agent Runtime Data Schema & Taxonomy
The Agent Runtime organizes data through a structured pipeline to ensure consistency across Openclaw Skills implementations:
| Component | Responsibility | Data Type |
|---|---|---|
| ToolRegistry | Tool registration, search, and routing | Object Map |
| PermissionPolicy | Permission level control and security | Policy Config |
| HookRunner | Intercepting PreToolUse and PostToolUse | Middleware |
| SessionCompactor | Context and history compression | String/Buffer |
| UsageTracker | Token and usage statistics tracking | JSON Analytics |
Agent Runtime Advanced Features
- Support for multiple specialized agent types including explore, plan, and verify.
- Dynamic tool registration allowing agents to expand their capabilities on the fly.
- Advanced hook system for injecting custom logic before and after tool execution.
- Integrated context compactor to maintain performance during long-form interactions.
- Detailed usage tracking for monitoring costs and token efficiency within Openclaw Skills.
SKILL.md
Loading
Related Openclaw Skills
Jike Digest
An automated content curation tool that extracts the latest high-value posts from Jike topics and transforms them into structured Markdown digests.
Usage Tracker
A comprehensive token monitoring and budget management system designed to track AI agent consumption and operational costs.
Tool Registry
A sophisticated tool discovery and registration system featuring token-based routing, granular permissions, and sub-agent whitelisting.
Hook System
A powerful tool interception framework to inject custom logic before and after AI tool execution.
MAL Anime Tracker
A command-line tool for managing MyAnimeList watchlists and receiving automated notifications through Openclaw Skills.
Dandan Test Skill
A diagnostic utility for verifying the publishing lifecycle and CLI functionality of AI agent skills.
Featured*
