smart-wake

What This Skill Does

The smart-wake skill is a critical architectural utility for OpenClaw agents designed to handle long-running, multi-stage, or high-latency tasks that would otherwise fail due to subagent timeout constraints. By implementing a standardized checkpointing system, this skill allows an agent to serialize its current progress, persist it to memory, and schedule a resurrection via the Gateway cron mechanism. When a task approaches its execution limit, smart-wake performs a graceful shutdown, saves the state to a specific task_id schema, and registers an auto-wake signal. Upon restarting, the agent consumes the checkpoint, verifies the integrity of previous artifacts, and resumes exactly where it left off, ensuring full idempotency. It transforms fragile, timeout-prone workflows into robust, self-healing processes.

Installation

To install this skill, use the following command in your OpenClaw environment: clawhub install openclaw/skills/skills/cong91/smart-wake

Use Cases

• Deep Data Crawling: Scrapers that need to parse thousands of records across multiple pages that exceed typical session duration limits.
• Automated Research: Multi-step analytical agents that aggregate information from many sources and compile comprehensive reports.
• Overnight Processing: Tasks that need to run continuously during off-hours, requiring automatic continuation if the agent becomes unresponsive.
• Complex Pipeline Automation: Sequential workflows involving file manipulation or external service interaction that must pause and resume without losing session-specific state.

Example Prompts

1. "Perform a deep scan of the provided repository documentation, and use smart-wake to ensure the indexing process finishes even if it takes over an hour."
2. "Start a systematic audit of the database. Implement smart-wake so that if the task reaches the timeout limit, it saves the current progress and continues from the last user entry."
3. "Run the long-form analysis on this dataset and set up a smart-wake checkpoint every 10 minutes to guarantee we don't lose progress during the batch processing."

Tips & Limitations

• Granularity is Key: Break your tasks into 5-15 minute chunks. Smaller chunks ensure a higher success rate for checkpoints and faster recovery times.
• Idempotency Guard: Always implement checks in your code to verify if an artifact exists before attempting to write it again; never assume a previous task finished completely.
• Resource Management: While smart-wake handles timeouts, ensure that the Gateway cron interval is long enough to prevent an infinite loop of rapid session spawns that could overwhelm your system resources.
• Monitoring: Always inspect the errors array in your checkpoint schema after a resume to identify why the previous session reached the timeout threshold.