workflow-cache

What This Skill Does

The workflow-cache skill is a powerful performance and cost-optimization layer for OpenClaw agents. It functions as a crowdsourced registry of successful task execution patterns. Instead of requiring the LLM to "reason" and "explore" through every step of a browser-based task, the skill intercepts the user intent, queries the cloud-based registry for a pre-validated workflow, and executes the sequence immediately if a match is found. This bypasses costly inference cycles, reducing token consumption by up to 90%. By caching the output of successful automation runs, the system effectively turns trial-and-error sequences into permanent, repeatable shortcuts. It includes a PII Sanitizer to ensure that sensitive user data is stripped before any workflow patterns are uploaded, maintaining high privacy standards while contributing to the global knowledge base.

Installation

To install this skill, use the ClawHub CLI tool in your terminal:

clawhub install openclaw/skills/skills/ainclaw/workflow-cache

If you prefer a manual setup or need to customize the build, you can navigate to the local workspace:

cd ~/.qclaw/workspace/skills/workflow-cache
npm install
npm run build

Use Cases

• High-Frequency Data Extraction: Automating daily reports from dynamic web dashboards where the steps remain consistent but the token cost per run is traditionally high.
• Complex Multi-Step Forms: Navigating multi-page registration or checkout flows where the LLM might otherwise struggle with state management.
• Regression Testing: Rapidly verifying that specific UI elements or workflow paths are still functional across different agent sessions.
• Cost-Sensitive Environments: Scaling automation across an entire enterprise team without seeing a massive spike in API token usage.

Example Prompts

• "OpenClaw, scrape the latest sales figures from the dashboard and save to my local CSV using the standard workflow-cache pattern."
• "Navigate to the registration portal and complete the onboarding steps, leveraging cached workflows for maximum efficiency."
• "Perform the daily cloud infrastructure health check for all zones; please prioritize cached execution to minimize latency."

Tips & Limitations

• Enablement: Ensure auto_contribute is set to true if you want to help the community by sharing your successful workflow patterns.
• Caching Latency: The timeout_ms default is 300ms; if you are on a high-latency network, consider increasing this to ensure the cloud query has sufficient time to return a match.
• PII Awareness: While the skill includes a PII Sanitizer, always be cautious about what sensitive inputs are passed through the automation path. The skill caches logic (selectors and clicks), not the data content.
• Website Changes: When a website updates its DOM structure, cached workflows might fail. The system handles this by flagging outdated workflows for update, but keep an eye on edge cases for major platform redesigns.