Best-practice guides
From Agent frameworks to AI coding clients — step-by-step recipes to get Starsg running.
OpenClaw — Agent framework integration
Use Starsg as the unified inference backend for the OpenClaw Agent framework. Five minutes to integrate.
CLI setup — Claude Code / Codex
Point the Claude Code and Codex CLIs at Starsg so every terminal-based assistant shares one gateway and one billing view.
Prompt engineering best practices
From role setup to iteration — write prompts that are reusable, evaluable and not just one-off magic.
OpenCode — one config, three providers
Point OpenCode at Starsg and use Claude / GPT / Gemini side by side from one terminal, sharing one key and one usage view.
AI Coding — Cursor / Continue / Cline
Plug Starsg into popular AI coding clients (Cursor, Continue, Cline) and manage every model key and quota in one place.
RAG — retrieval-augmented generation
Combine Embedding + Reranker + Command R+ to power a high-quality enterprise knowledge-base Q&A.
Multi-agent collaborative workflows
Orchestrate Claude / GPT and other models together to deliver complex tasks at the lowest cost.
Function calling (Tool Use) guide
Structured outputs + Tool Use — how to call tools reliably on Starsg.
Multimodal — images and document understanding
GPT-5o / Gemini 2.5 multimodal templates and image-token billing notes.