skills.rest
decniner
decninerCOMMUNITY
Explore all skills by this author →
GitHub

claude-proxy-backends

Route Claude Code CLI through local proxies to alternative LLM providers.

Updated Jul 3, 2026
Search Tags:#claude code#llm routing#openrouter#proxy#ollama#free tier#api translation
One-click install
npx skills add https://github.com/decniner/HermesP1 --skill claude-proxy-backends-decniner
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: claude-proxy-backends
Source: https://github.com/decniner/HermesP1/tree/main/.hermes-backup/skills/autonomous-ai-agents/claude-proxy-backends
Command: npx skills add https://github.com/decniner/HermesP1 --skill claude-proxy-backends-decniner

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve? Claude Code CLI normally requires an Anthropic subscription and only works with Anthropic's native API. This Skill lets you route Claude Code through a local translation proxy so you can use alternative LLM providers like OpenRouter, Ollama, or vLLM, enabling free-tier usage, cost optimization, local/air-gapped operation, and model testing. ## Core Features & Use Cases - Proxy-Based Provider Routing: Configure the community claude-code-proxy (FastAPI) to translate Anthropic message/tool format into OpenAI-compatible Chat Completions for any function-calling provider. - Client-Side Validation Workaround: Set ANTHROPIC_MODEL to a spoofed valid model name to bypass Claude Code v2.1+ hardcoded model validation before requests reach the proxy. - Provider-Specific Configs: Ready-made .env setups for OpenRouter free tier (with auto-routing via openrouter/free), Ollama local models, and vLLM endpoints. - Use Case: Run Claude Code workflows in CI or on a machine without Anthropic billing by pointing ANTHROPIC_BASE_URL at a local proxy backed by OpenRouter free-tier models. ## Quick Start Set up the claude-code-proxy with an OpenRouter API key and configure my shell environment so Claude Code routes through the local proxy on port 8082.

Frequently Asked Questions about claude-proxy-backends

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

FAQPage Schema
How do I use Claude Code with OpenRouter free models?▼

Run the fuergaosi233/claude-code-proxy locally with OPENAI_BASE_URL set to https://openrouter.ai/api/v1 and your OpenRouter key in .env. Then point Claude Code at the proxy via ANTHROPIC_BASE_URL=http://127.0.0.1:8082 and set ANTHROPIC_MODEL to a valid Anthropic model name.

How to run Claude Code with local models via Ollama?▼

Set the proxy's OPENAI_BASE_URL to http://localhost:11434/v1 and configure BIG_MODEL to a local model like qwen2.5-coder:32b. Ollama 0.3.0+ serves an OpenAI-compatible endpoint, and the proxy translates Claude Code's Anthropic-format requests to it.

Why does Claude Code reject my custom model name?▼

Claude Code v2.1+ validates model names client-side against a hardcoded list in its compiled binary before sending any request. Set ANTHROPIC_MODEL to a known-valid name like claude-3-5-sonnet-20241022 so the request passes validation and reaches the proxy.

Does the Claude Code proxy support MCP servers and plugins?▼

No. The proxy only translates the core messages and tools API between Anthropic and OpenAI formats. MCP server integration, plugin systems, OAuth login, and session persistence features are unavailable through the proxy.

Why do Claude Code proxy requests time out on free tiers?▼

OpenRouter free-tier models are aggressively rate-limited and slow, and the proxy has a 90s request timeout. Large outputs over roughly 10KB regularly time out, so split work into small sub-tasks and prefer openrouter/free for auto-routing to less congested backends.

Related Skills

Explore More Software Engineering Skills→
inho-teaminho-team
6

Qlocal-llm-setup

Configure local LLM servers and integrate them with Claude Code.

Official
Advanced
jleechanorgjleechanorg

claude-code

Delegates coding tasks to Claude Code CLI for code generation, review, and PR workflows.

Official
Advanced
adiytharpansaadiytharpansa

freeride

Discover, rank, and rotate free OpenRouter models for OpenClaw configurations.

Community
Advanced
chiptoe-svgchiptoe-svg

add-codex

Configure OpenAI Codex as a NanoClaw agent provider with flexible authentication.

Community
Intermediate