AI Agent Hub
Back to plugins
🧩

dsh-rate-limiter

admin-security Updated 2026.08.25

Run the following command in DeepSeek Harness:

dsh plugin install Xidong-AI/dsh-rate-limiter

Paste the following prompt into your AI chat to install this plugin:

To install this plugin in DeepSeek Harness, run: dsh plugin install Xidong-AI/dsh-rate-limiter. Source: https://github.com/Xidong-AI/dsh-rate-limiter

About this plugin

When building applications with DeepSeek Harness, developers often face frequent HTTP 429 "Too Many Requests" errors from upstream APIs due to request overload, which disrupts workflows and harms system stability. The dsh-rate-limiter plugin proactively controls request rates before they are sent, employing a per-provider token bucket algorithm to set request caps, thereby preventing 429 errors at their source. Its core capability lies in intelligently delaying over-limit requests instead of rejecting them, ensuring all requests are queued and processed without loss. It perfectly complements the official dsh-llm-retry plugin: rate limiting acts as prevention before requests, while retry provides a safety net after failures, with both plugins coexisting seamlessly. This plugin is ideal for development teams and operations personnel requiring fine-grained API call management, high reliability, and automated recovery mechanisms.

Use Cases

  • Controlling request rates to multiple AI providers during peak hours to prevent service interruptions.
  • Ensuring smooth handling of requests when fine-tuning API quotas for different providers.
  • Integrating into existing DeepSeek Harness workflows to achieve high reliability alongside retry plugins.

Best For

  • Developers needing reliable management of API call frequencies.
  • Operations teams responsible for optimizing service stability and resource usage.
  • Project teams building complex applications based on DeepSeek Harness.