bifrost-gatewaylisted

Routes EVERY LLM / AI-provider request through a Bifrost (OpenAI-compatible) gateway so that all calls are logged and tracked in one place. Use this skill WHENEVER you write, configure, or debug any code that calls a large language model or AI provider — OpenAI/GPT, Anthropic/Claude, Google Gemini, DeepSeek, xAI/Grok, Perplexity, or any other — in ANY language (Python, PHP, JavaScript/TypeScript, Go, curl) or framework (Laravel AI SDK, LangChain, LlamaIndex, Vercel AI SDK). This covers new integrations, editing existing client setup, base_url / api_key configuration, chat completions, the Responses API, embeddings, and streaming. Trigger even when the user only says "call gpt-4o", "add Claude to this script", "sk-bf-...", "generate text with an LLM", or names a provider SDK without mentioning the gateway. NEVER call provider APIs directly — the whole point of the gateway is that no request goes around it.
SiteVisorCloud/bifrost-gateway-integration-skill · ★ 1 · AI & Automation · score 72

Install: claude install-skill SiteVisorCloud/bifrost-gateway-integration-skill

# Bifrost AI Gateway There is a **Bifrost** LLM gateway (OpenAI-compatible) at `https://your-gateway.example.com`. Every call to every AI provider must go through it, so that all requests are logged and tracked in one place. Hitting `api.openai.com`, `api.anthropic.com`, `generativelanguage.googleapis.com`, etc. directly defeats the whole purpose: those requests would be invisible to the gateway's logging. So the job of this skill is simple: **the moment you touch code that talks to an LLM, point it at the gateway instead of the provider.** In practice that's almost always a one-line change (the `base_url`) plus using the gateway key. > Replace `your-gateway.example.com` with your own gateway host (set it once via > the `BIFROST_BASE_URL` env var). Everything else stays identical. ## No gateway yet? — Bifrost quick start (one-time) If no gateway is running, stand one up with Bifrost (the open-source LLM gateway): ```bash docker run -p 8080:8080 -v "$(pwd)/data:/app/data" maximhq/bifrost # or, without Docker: npx -y @maximhq/bifrost ``` Open `http://localhost:8080`, add your providers' API keys, and create a virtual key under Governance (that key — commonly prefixed `sk-bf-` — is your `BIFROST_API_KEY`). Then `BIFROST_BASE_URL` is `http://localhost:8080/v1` (or your deployed host + `/v1`). Full steps: [references/install-bifrost.md](references/install-bifrost.md). ## The three things you always need 1. **Base URL:** `https://your-gateway.example.com/v1` 2. **Model st