Integration Overview - Hostlinku Docs

Almost all modern AI coding agents and IDEs can be configured to use other providers via a custom base URL. Since Hostlinku supports three protocols at once—OpenAI Chat Completions, OpenAI Responses, and Anthropic Messages—you can use a single API key and a single balance for all the tools listed below, then monitor all usage from a single Request Logs view in the console.

Choose your tool

Tool	API Format Used	Base URL	Guide
Claude Code (CLI & VS Code/JetBrains extensions)	Anthropic Messages	`https://api.hostlinku.com`	Claude Code
Codex CLI	OpenAI Responses	`https://api.hostlinku.com/v1`	Codex CLI
Cursor	OpenAI Chat Completions	`https://api.hostlinku.com/v1`	Cursor
VS Code (built-in Chat & Agent modes)	OpenAI Chat Completions (or Anthropic/Responses)	`https://api.hostlinku.com/v1/chat/completions` (full endpoint URL)	VS Code
GitHub Copilot CLI & Copilot app	OpenAI Chat Completions (or Anthropic)	`https://api.hostlinku.com/v1`	GitHub Copilot
Zed (Agent Panel)	OpenAI Chat Completions	`https://api.hostlinku.com/v1`	Zed
Cline, Roo Code	OpenAI Chat Completions (or Anthropic)	`https://api.hostlinku.com/v1`	Cline & Roo Code
Kilo Code	OpenAI Chat Completions (or Anthropic)	`https://api.hostlinku.com/v1`	Kilo Code
Continue (VS Code & JetBrains)	OpenAI Chat Completions	`https://api.hostlinku.com/v1`	Continue
JetBrains AI Assistant (IntelliJ IDEA, PyCharm, WebStorm, …)	OpenAI Chat Completions	`https://api.hostlinku.com/v1`	JetBrains
Junie CLI (JetBrains)	OpenAI Chat Completions (or Anthropic)	`https://api.hostlinku.com/v1/chat/completions` (full endpoint URL)	Junie
Factory Droid (CLI & desktop)	Anthropic Messages or OpenAI Chat Completions	`https://api.hostlinku.com` / `https://api.hostlinku.com/v1`	Factory
OpenCode	OpenAI Chat Completions	`https://api.hostlinku.com/v1`	OpenCode
Xum (parallel agent)	OpenAI Chat Completions (or Anthropic)	`https://api.hostlinku.com/v1`	Xum
Aider	OpenAI Chat Completions	`https://api.hostlinku.com/v1`	Aider
OpenAI Agents SDK (Python & TypeScript)	OpenAI Responses	`https://api.hostlinku.com/v1`	Agents SDK

Other tools that offer “OpenAI-compatible” or “custom base URL” options (e.g., Open WebUI, LibreChat, LangChain, Vercel AI SDK, n8n, Dify) follow the same pattern: base URL https://api.hostlinku.com/v1, API key sk-…, and a model slug from the catalog.

Three Things That Always Stay the Same

Base URL — https://api.hostlinku.com/v1 for OpenAI-formatted tools; https://api.hostlinku.com (without /v1) for Anthropic-formatted tools, since the Anthropic SDK adds /v1/messages on its own.
API key — the sk-… secret from console.hostlinku.com → API Keys. Used as Authorization: Bearer sk-… (OpenAI) or x-api-key (Anthropic); the gateway accepts both at all /v1/* endpoints.
Model slug — exactly as listed in the model catalog, e.g., claude-sonnet-4-6, gpt-5-5, deepseek-v4-flash. Each API format can call any model—for example, Claude via the Codex CLI, or GPT via Claude Code—because the gateway translates the protocols.

Recommended models for coding

A list of active slugs for your account is always available in the console (under the Model menu) or at GET /v1/models. To get started:

Requirement	Slug	Notes
Daily agentic coding (safe default)	`claude-sonnet-4-6`	Powerful tool calling, large context.
Difficult tasks / cross-file refactoring	`claude-opus-4-6`, `gpt-5-5`	More expensive; use when Sonnet isn’t enough.
Codex CLI	`gpt-5-3-codex`, `gpt-5-5`	Optimized for Codex workflows (apply_patch, shell).
Fast & inexpensive models for sub-agents / commit messages / summaries	`claude-haiku-4-5`, `deepseek-v4-flash`	Suitable for haiku aliases / weak models.
Large-scale open-weight alternatives	`kimi-k2-7-code`, `glm-5-3`, `qwen3-6-plus`	All support tool calling.

Safe Practices

The API key is only valid for the /v1/* gateway. The sk-… key cannot be used to log in console, change your balance, or create a new key—so the key stored in the IDE’s configuration cannot take over your account.
Do not commit your API key. Store it in an environment variable (HOSTLINKU_API_KEY) or as a secret their respective store tools; all the guides here follow that pattern.
Monitor usage. The Request Log menu in the console displays every request from each tool — model, token, cost, latency, and API key used. See Usage & Limits.

Common Issues

Symptoms	Causes & Solutions
`401` / authentication error	The API key is incorrect, revoked, or expired. Make sure no spaces or newlines were copied along with it. In Claude Code, run `/logout` first if you were previously logged in with an Anthropic account.
`402` / insufficient quota	Balance is low or the API key credit limit has been reached. Top up your balance or increase the limit in the console.
`404` / model not found	The model slug is misspelled or does not exist in the catalog. Copy it exactly as it appears from `GET /v1/models`.
`400` not available in Service Mode	The model is not available in the Service Mode of that API key. Change the model or create a key with a different Service Mode.
`429` / rate limit	The tool is sending too many parallel requests (sub-agents). Please wait for the time specified in the `Retry-After` header, or reduce the number of parallel requests. See Usage & Limits.
The tool rejects the base URL `localhost`	Some tools (e.g., Cursor) call the API from their own servers, not from your computer. `https://api.hostlinku.com` is a public endpoint, so it’s safe.
The autocomplete / Tab feature isn’t working	Inline autocomplete typically uses the `/v1/completions` endpoint (FIM) or the tool’s proprietary infrastructure. Hostlinku handles chat, Responses, and Messages—not FIM. Use Hostlinku for chat/agent/edit; leave autocomplete to the tool’s built-in provider.

Need another tool that isn’t listed here? Send it to halo@hostlinku.com or check out Support.

​Choose your tool

​Three Things That Always Stay the Same

​Recommended models for coding

​Safe Practices

​Common Issues

Choose your tool

Three Things That Always Stay the Same

Recommended models for coding

Safe Practices

Common Issues