A comprehensive framework for creating, configuring, and orchestrating autonomous AI agents that think and evolve independently.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install autonomous-agent-toolkit
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 autonomous-agent-toolkit 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 Autonomous Agent Toolkit?
The Autonomous Agent Toolkit is a specialized suite within Openclaw Skills designed to transform static AI interactions into persistent, self-driven entities. Developed by Forge, the toolkit provides the architectural blueprint for building agents capable of operating on schedules, maintaining long-term memory, and executing complex workflows without constant human intervention.
At its core, this skill leverages a file-based persistence model, ensuring that an agent's identity, goals, and learnings are preserved across sessions. By utilizing structured markdown files and automated cron jobs, users can deploy AI solopreneurs or multi-agent teams that handle everything from routine monitoring to strategic decision-making.
Autonomous Agent Toolkit Use Cases
- Creating a new autonomous agent persona with a defined SOUL.md identity.
- Setting up multi-agent teams where specialized roles like Builder and Marketer collaborate through shared workspaces.
- Configuring automated nightly reviews to distill daily logs into long-term MEMORY.md insights.
- Implementing heartbeat schedules for periodic task execution and autonomous awareness checks.
- Building self-improving systems that analyze past performance to optimize future decision-making.
How Autonomous Agent Toolkit Works
- Initialize the agent using the toolkit's generator script to create a standardized workspace structure.
- Define the core identity and decision-making framework in the SOUL.md and AGENTS.md files.
- Establish a memory protocol by setting up a directory for daily logs and a primary long-term memory file.
- Configure Openclaw Skills cron jobs to trigger autonomous heartbeats and maintenance tasks at specific intervals.
- Orchestrate multi-agent interactions by defining shared file-based communication channels and specific model routing based on task complexity.
- Monitor agent performance through the audit trail and daily logs, iterating on the SOUL.md rules as needed.
Autonomous Agent Toolkit Setup
To begin building with these Openclaw Skills, run the following commands to generate your agent structure:
# Generate a new agent directory
python3 scripts/generate_agent.py --name "MyAgent" --path ./my-agent
# Add a heartbeat cron job for autonomous operation
openclaw cron add --name "Heartbeat" --cron "*/20 6-22 * * *" --message "Check HEARTBEAT.md and execute tasks."
Autonomous Agent Toolkit Data Schema & Taxonomy
The toolkit organizes agent data through a strict file-based taxonomy to ensure consistency across the Openclaw Skills ecosystem:
| File / Directory | Description |
|---|---|
SOUL.md |
The master identity file: voice, goals, and the priority stack for decision-making. |
MEMORY.md |
Curated long-term knowledge and distilled lessons learned over time. |
HEARTBEAT.md |
The queue for pending autonomous tasks and periodic checks. |
memory/ |
A directory containing append-only daily logs named by date (YYYY-MM-DD.md). |
status/ |
JSON files tracking the health and activity state of various agents in a team. |
Autonomous Agent Toolkit Advanced Features
- Multi-agent orchestration using model routing (e.g., Opus for strategy, Haiku for monitoring).
- Automated nightly self-improvement loops that update long-term memory systems.
- File-based inter-agent communication for collaborative multi-agent workflows.
- Priority-based decision frameworks that override default LLM behaviors with specific business logic.
- Built-in safety guardrails including hard red lines and human-in-the-loop triggers.
SKILL.md
Loading
Related Openclaw Skills
Juya AI Daily Fetcher
An automated skill to fetch, parse, and format the Juya AI Daily newsletter for quick consumption of AI industry updates.
AI Intelligent Contract Review
Automate contract review, risk identification, and clause analysis with high-precision AI models.
Beijing Morning Weather Assistant
An automated morning briefing skill that provides Beijing weather updates along with specific lifestyle recommendations for clothing and nutrition.
Coding
A persistent memory skill that captures and applies your specific coding preferences and stack decisions through explicit user feedback.
A/B Test Setup
A comprehensive framework for designing, implementing, and analyzing statistically rigorous A/B tests to drive data-informed product decisions.
Desk Research Skill
A high-precision research framework for AI agents to conduct structured secondary research, market landscape scans, and evidence-based synthesis.
Featured*
