gem-llm-gateway-debug

Diagnose GEM-LLM Gateway authentication, rate limiting, and proxy routing failures.

Updated Apr 4, 2026
One-click install
npx skills add https://github.com/saintgo7/claude-skills --skill gem-llm-gateway-debug
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: gem-llm-gateway-debug
Source: https://github.com/saintgo7/claude-skills/tree/main/gem-llm-gateway-debug
Command: npx skills add https://github.com/saintgo7/claude-skills --skill gem-llm-gateway-debug

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

GEM-LLM Gateway issues like 401 invalid API keys, unknown_model errors, rate limiting, missing /v1/chat endpoints, and broken streaming responses are hard to diagnose and often caused by misconfigured middleware, environment variables, or upstream model mappings.

Core Features & Use Cases

  • Endpoint-first diagnostics: Validates health, readiness, metrics, and model list behavior so you can localize failures quickly.
  • Auth, quota, and proxy troubleshooting: Covers API key hashing/salt, admin key management, RPM/concurrency/daily token quota checks, and vLLM upstream routing.
  • Streaming reliability checks: Ensures SSE chunk pass-through and proper [DONE] termination while keeping usage logging non-blocking.
  • Practical fixes for common failures: Provides specific causes for 401/400/429/500/502 and links remediation paths (e.g., config updates, SQLite WAL recovery, proxy upstream issues).

Quick Start

Run the skill install command: ./install.sh gem-llm-gateway-debug.

Frequently Asked Questions about gem-llm-gateway-debug

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

FAQPage Schema
Why does my FastAPI gateway return 401 invalid API key errors?

A 401 invalid API key error occurs when the gateway authentication middleware rejects a request due to misconfigured API key hashing or salt values. Validating Bearer key flows and checking admin key management configurations will pinpoint and resolve the authentication failure.

How do I fix unknown_model mapping mismatches in a vLLM proxy?

To fix unknown_model mapping mismatches in a vLLM proxy, you must correct the upstream routing configurations that map requested model names to their specific vLLM upstreams. Diagnosing the model list behavior helps localize these misconfigurations quickly.

What causes broken streaming SSE pass-through under /v1/chat/completions?

Broken streaming SSE pass-through under /v1/chat/completions is caused by disrupted SSE chunk transmission or improper [DONE] termination. You need to perform streaming reliability checks to ensure proper chunk pass-through while keeping async usage logging non-blocking.

How do I troubleshoot Gateway 500 responses and QueuePool saturation?

Troubleshooting Gateway 500 responses and QueuePool saturation requires running endpoint-first diagnostics on health and readiness endpoints. These failures often point to specific remediation paths like applying SQLite WAL recovery or resolving underlying proxy upstream connection issues.

How do I validate RPM and daily token quota enforcement in an API gateway?

To validate RPM, concurrency, and daily token quota enforcement in an API gateway, you must inspect the rate limiting and quota enforcement middleware configurations. Checking these quota parameters ensures the gateway properly restricts traffic according to defined limits.

Why are my Prometheus metrics missing or incorrect for the API gateway?

Missing or incorrect Prometheus metrics occur when the gateway fails to properly expose or record monitoring data. Validating the metrics endpoint behavior and ensuring async usage logs record without blocking will restore accurate metric reporting.