autonomous-tasks

What This Skill Does

The autonomous-tasks skill transforms your OpenClaw agent into a persistent, self-driven worker. It operates on a robust 'Read-Generate-Execute-Log' loop that allows the agent to handle complex, multi-step projects without constant manual intervention. When triggered, the agent reads your defined long-term goals, breaks them down into manageable, actionable tasks, executes them sequentially, and maintains a clear audit trail of completed work, failures, and new ideas found along the way.

Installation

To add this capability to your agent, execute the following command in your terminal: clawhub install openclaw/skills/skills/changye01/autonomous-tasks

After installation, initialize the agent by allowing it to create the agents/ directory structure. We strongly recommend setting up a cron job for consistent progress: openclaw cron add --name "autonomous-tasks" --message "run autonomous tasks" --every 1h.

Use Cases

• Long-term Research: Task the agent with researching a specific technical topic over several days, gathering links and summarizing findings into files.
• Content Pipeline: Configure the agent to generate blog post drafts, refine them, and save them to your local project folder periodically.
• Project Maintenance: Automate recurring development tasks such as dependency auditing or documentation updates.
• Task Decomposition: Use the agent to take a high-level project roadmap and turn it into actionable daily milestones.

Example Prompts

1. "Set goal: Research the latest updates in the Rust language and summarize the top three features for our dev team in rust-updates.md."
2. "New goal: Organize my download folder and archive all images from the last month into a subfolder named 'Old_Assets'."
3. "Run goals: Check the backlog and pick up the next task related to the project documentation."

Tips & Limitations

• Granularity: The agent works best when tasks are specific. If the agent struggles, break your milestones into smaller steps.
• State Management: All progress is stored in agents/. Do not move or delete this folder unless you intend to reset the agent's memory.
• Safety: Always review memory/tasks-log.md periodically to ensure the agent is heading in the right direction. It will not retry failed tasks automatically, so you may need to intervene if critical tasks are marked with [!].