An autonomous evolution engine that balances internal self-optimization with external value creation and automated error recovery.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install auto-evolver
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 auto-evolver 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 auto-evolver?
auto-evolver is a sophisticated execution system designed for AI agents to achieve continuous growth through a dual-loop mechanism. It distinguishes itself within Openclaw Skills by ensuring that AI cycles result in actual output rather than empty processing loops. By alternating between an Inner Loop for self-diagnosis and an Outer Loop for value-driven connections, the system maintains a high standard of operational integrity.
The engine is built on the principle that true evolution requires tangible production. It includes built-in safeguards such as token budget management, idle detection to prevent wasted resources, and an automated repair system that learns from recurring failures. This makes it an essential tool for developers looking to build resilient, self-improving autonomous agents.
auto-evolver Use Cases
- Autonomous system health monitoring and self-healing for long-running AI agents.
- Automated content publishing and engagement tracking across platforms like Zhihu and Gist.
- Proactive task generation by monitoring market trends, research papers, and repository growth.
- Resource-constrained environment management via strict token budget and idle protection protocols.
How auto-evolver Works
- The Inner Loop runs every 5 minutes to perform system diagnostics, check token usage (keeping single cycles <8000 tokens), and scan for recurring failures.
- If failures exceed thresholds, the system automatically generates repair tasks in the pending queue.
- The Outer Loop runs every 15 minutes to execute value-generating tasks such as publishing content or conducting research.
- Connection tracking monitors feedback (likes, replies, conversions) and logs high-quality interactions in the connection log.
- An idle detection mechanism monitors output; if no productive output is detected for 3 consecutive cycles, the system pauses to save resources.
- Task generation agents scan external sources like Web Search and TrustMRR daily to populate the task queue with high-value opportunities.
auto-evolver Setup
To deploy this skill within your Openclaw environment, use the following commands to initialize the dual-loop cron jobs:
# Add the Inner Loop (Self-Optimization) to cron every 5 minutes
openclaw cron add "*/5 * * * *" "auto-evolver inner-loop"
# Add the Outer Loop (Value Connection) to cron every 15 minutes
openclaw cron add "*/15 * * * *" "auto-evolver outer-loop"
# Manually verify the system status
openclaw skill run auto-evolver --action status
auto-evolver Data Schema & Taxonomy
The auto-evolver skill organizes its operational data into structured Markdown logs to ensure transparency and traceability:
| Data Type | Path | Description |
|---|---|---|
| Evolution Logs | memory/auto-evolver-YYYY-MM-DD.md |
Records of diagnostic results, token usage, and task generation. |
| Connection Logs | memory/connection-log-YYYY-MM-DD.md |
Tracks external platform engagement, interaction quality, and conversion rates. |
| Task Queue | inbox/grok-tasks/TASK-YYYY-MM-DD-XXX.md |
Individual task files generated by the autonomous discovery engine. |
| Alerts | inbox/grok-tasks/IDLE-ALERT-YYYY-MM-DD-HHMM.md |
Emergency notifications triggered by idle protection or budget overruns. |
auto-evolver Advanced Features
- Multi-source autonomous task generation (Foundry Insights, Web Search, TrustMRR, 4D Compression analysis).
- Automated repair hooks for common environment errors like browser relay issues or file system ENOENT errors.
- Dynamic resource scaling that degrades functionality gracefully based on daily and weekly token quota consumption.
- Connection quality protection that filters out low-value interactions from performance statistics to ensure data integrity.
SKILL.md
Loading
Related Openclaw Skills
Data Analytics Assistant
An intelligent data processing agent that automates website traffic monitoring, user behavior analysis, and conversion rate optimization.
Connection Tracker
An automated tracking dashboard that monitors and evaluates the value of all external and internal connections established by an AI agent.
Email Helper
An intelligent AI assistant for drafting professional emails, managing templates, and optimizing inbox organization.
Ellya: Virtual AI Companion & Synthetic Media Generator
Ellya is a sophisticated virtual companion skill designed to bootstrap autonomous personalities and generate highly personalized synthetic media and photo series.
Munger Decision Assistant
A structured decision-making assistant that applies Charlie Munger's mental models to analyze complex investment and strategy scenarios.
C-Task-Cleanup
An automated maintenance utility that cleans completed Class-C tasks and extracts high-value insights into long-term memory.
Featured*
