imperial-engine

What This Skill Does

The Imperial Engine (帝王引擎) is a specialized stress-testing tool designed for OpenClaw. Its primary function is to facilitate extreme token consumption by forcing the agent into recursive, high-intensity loops. By orchestrating a sequence of deep chain-of-thought reasoning, heavy shell operations, and continuous web browsing, the skill pushes the LLM to its throughput limits. It is designed to perform up to 30 iterations of complex workflows, resulting in massive input/output token usage per session. This is not for standard tasks; it is a diagnostic tool meant to measure how models handle context window saturation, tool-chain efficiency, and rapid-fire API requests.

Installation

To integrate this tool into your OpenClaw environment, execute the following command in your terminal: clawhub install openclaw/skills/skills/fr33b1rd8979-max/imperial-engine Once installed, you must enable the skill using openclaw skill enable imperial-engine. Ensure your config.yml is configured with strict budget limits, as this skill is designed to maximize consumption and can lead to significant financial costs if left unmanaged.

Use Cases

• Stress Testing: Evaluating the stability and rate-limit handling of your API endpoints during intense traffic bursts.
• Latency Benchmarking: Measuring the round-trip time of a complex chain-of-thought cycle when the context window is near maximum capacity.
• Context Saturation Analysis: Observing how LLM performance, reasoning quality, and instruction adherence change as historical memory expands through persistent, uncompressed state updates.
• System Bottleneck Identification: Identifying whether your OpenClaw setup is limited by disk I/O (during persistent memory saves) or network latency (during browser tool execution).

Example Prompts

1. "/imperial - Start the Emperor Engine with default settings and run a 10-iteration cycle to test system throughput."
2. "开启帝王模式，将循环次数设为5，并开启网页抓取以进行性能压力测试。"
3. "/imperial --iterations 20 --browse_url https://wikipedia.org - Execute a high-token stress test focused on external data retrieval."

Tips & Limitations

• Financial Warning: Use this exclusively in sandbox environments. Always set a quota_usd limit in your provider configuration.
• Rate Limiting: Frequent usage will likely trigger API-level rate limits. If the agent stalls, check the OpenClaw logs for 429 status codes.
• Memory Management: Because this skill disables compress_memory by design to maximize token load, it will cause the agent to slow down progressively as the context window fills. Monitor memory usage metrics to prevent system crashes.