model-dos

Enforce token limits, throttling, and timeouts on LLM API endpoints.

20|Updated Feb 27, 2026
One-click install
npx skills add https://github.com/thejefflarson/soundcheck --skill model-dos
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: model-dos
Source: https://github.com/thejefflarson/soundcheck/tree/main/.claude/skills/model-dos
Command: npx skills add https://github.com/thejefflarson/soundcheck --skill model-dos

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Protects LLM endpoints from resource exhaustion by enforcing input caps, per-user throttling, and strict timeouts.

Core Features & Use Cases

  • Input length caps and token limits to bound prompts before they reach the model.
  • Per-identifier throttling to prevent abuse from a single user or API key.
  • Explicit inference deadlines to ensure timely responses even when upstreams misbehave.
  • Applicable to any service exposing an LLM API, including chatbots and multi-turn workflows.

Quick Start

Configure your LLM endpoint to enforce max tokens, rate limits, and request timeouts for incoming prompts.

Frequently Asked Questions about model-dos

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I prevent LLM API denial of service from unbounded prompts?

Prevent LLM DoS by enforcing input length caps, max token limits, and strict request timeouts to block resource exhaustion before prompts reach the model.

What is per-identifier throttling for LLM endpoints?

Per-identifier throttling limits API requests from a single user or key, preventing chat handlers and multi-turn workflows from triggering resource exhaustion through rapid abusive calls.

How do I cap prompt history in multi-turn chatbot workflows?

Cap prompt history by bounding conversation length inputs before they reach the model, ensuring inferences remain bounded and timely even when upstream calls misbehave.

Does this rate-limiting approach work with any LLM service?

Yes, this approach applies to any service exposing an LLM API. It enforces max tokens, input caps, and timeouts across chat handlers and multi-turn workflows.

Why do my LLM endpoints hang during slow upstream calls?

Endpoints hang because upstream calls lack explicit inference deadlines. Enforcing strict request timeouts guarantees timely responses and prevents resource exhaustion from slow upstreams.

When should I use prompt capping for API security?

Use prompt capping when exposing LLM APIs to prevent resource exhaustion. It bounds recursive prompts, enforces token limits, and applies per-user throttling to stop abuse.