Use AxForge with your tools

AxForge speaks the OpenAI API, and the Anthropic one too, so the tools you already use work here. Each guide below was tried for real on 2026-10-17: every step, with screenshots of what you will see.

Coding every day? AxForge Coder: the Coder plans give tokens every month, web search for your agent and cheaper nights.

Start here

Get your API keyThree steps in the console, and which model to pick. Every guide starts here.a minute

Coding agents in your editor

GitHub Copilot in VS CodeVS Code's own chat, with our models in its picker.VS Code ClineThe coding agent in the side bar, on the OpenAI Compatible provider.VS Code ContinueChat, edits and Agent mode, set up in its config.yaml.VS Code Kilo CodeThe coding agent with AxForge as a custom provider.VS Code

Coding agents in your terminal

Claude CodeAnthropic's agent, on our Messages API. One settings file.terminal Codex CLIOpenAI's agent, on our Responses API. One config file.terminal OpenCodeThe open-source agent, with AxForge as a provider.terminal aiderPair programming that commits each change. Two variables.terminal Qwen CodeThe Qwen team's agent. Three variables.terminal

For your coding agent

Web search for your agentOur own search as an MCP server: find pages on the web and in our docs, read one as text.MCP

Chat apps and automation

n8nWorkflows and AI agents, through its OpenAI credential.automation Open WebUIThe chat app you host yourself. One connection.self-hosted LibreChatThe open-source chat app, with a custom endpoint.self-hosted

In your code

OpenAI SDK and curlPython, Node or plain HTTP: the base URL changed, the rest the same.Python · Node Vercel AI SDKgenerateText and streamText on our models.TypeScript LangChainChatOpenAI with our base URL, for chains and agents.Python LlamaIndexOpenAILike, for retrieval and agents.Python LiteLLMThe proxy in front of many providers, or its Python SDK.proxy llmSimon Willison's command-line tool. One YAML entry.command line

The one pattern behind every guide

Whatever the tool, you tell it three things:

Set this	To
Base URL	https://api.axforge.ai/v1
API key	your AxForge key, from the console
Model	qwen3.8-27b, glm-5.3-flash, mimo-v2.6-flash or qwen3.8-flash-next

A tool that is not listed works the same way if it lets you set a custom OpenAI base URL. The quickest check that a key works, from any terminal:

curl https://api.axforge.ai/v1/chat/completions \
  -H "Authorization: Bearer $AXFORGE_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"qwen3.8-27b","messages":[{"role":"user","content":"ping"}]}'

$call = @{
  Method      = "Post"
  Uri         = "https://api.axforge.ai/v1/chat/completions"
  Headers     = @{ Authorization = "Bearer $env:AXFORGE_API_KEY" }
  ContentType = "application/json"
  Body        = '{"model":"qwen3.8-27b","messages":[{"role":"user","content":"ping"}]}'
}
(Invoke-RestMethod @call).choices[0].message.content

A reply with an answer means the key and the address are right.

When a tool will not talk to us: the connection doctor

One command that makes seven small checks on your key and says the fix for every cross: the key, the base URL, the model name, a short answer, a stream, a tool call, the MCP door. It makes a few tiny calls on your key (a handful of tokens) and sends nothing anywhere but our API. It asks for the key, or reads AXFORGE_API_KEY; set AXFORGE_BASE_URL to what you gave your tool and AXFORGE_MODEL to the model you use, and it checks those.

curl -fsSL https://dev.axforge.ai/assets/axforge-doctor.sh | bash

irm https://dev.axforge.ai/assets/axforge-doctor.ps1 | iex

Rather read it first? Download axforge-doctor.sh or axforge-doctor.ps1 (plain curl, plain PowerShell, no other tool) and run the file. Its exit code is the number of crosses, so a script can call it too.

Not here

Cursor takes a custom base URL only on its paid plans. Windsurf and Gemini CLI work with their own makers' models only. Roo Code shut down on 2026-05-15. Zed and JetBrains AI are not tried yet. Stuck with another tool? Talk to an engineer.

Ask on the forum Markdown For AI Updated 2026-11-06