Connect your coding agent

One command for all of them#

Run this in a terminal on macOS, Linux or Windows. It finds Claude Code, Codex, OpenCode, Kilo Code, Cline, Aider and Continue on your computer, signs you in through the browser, gives each tool its own key named after it, writes each tool's settings with a backup (including each model's context window, output limit, abilities and prices), and makes a test call through each one. Cline is set up for both its command line and its VS Code extension. For Roo Code and Cursor it shows the values to paste.

npx fullsendai connect

Later, npx fullsendai status shows each tool's spend this month, npx fullsendai model <id> switches every tool to another model, npx fullsendai doctor --fix repairs settings another update changed, npx fullsendai sync brings every tool up to date when models are added or renamed, and npx fullsendai disconnect puts every file back and revokes the keys. On a server or over SSH, add --key fs_live_… to skip the browser. The manual settings for each tool follow, if you'd rather set them yourself.

Which API to pick#

Claude Code#

Claude Code speaks Anthropic's API, so point it at FullsendAI with environment variables. The full Claude Code guide covers model aliases, per-project settings and troubleshooting.

export ANTHROPIC_BASE_URL=https://fullsendai.comexport ANTHROPIC_AUTH_TOKEN=$FULLSEND_API_KEYexport ANTHROPIC_MODEL=deepseek-v4.1-flashexport ANTHROPIC_DEFAULT_HAIKU_MODEL=deepseek-v4.1-flash   # the small model for background tasksclaude

Older Claude Code versions name the small model ANTHROPIC_SMALL_FAST_MODEL. ANTHROPIC_API_KEY works in place of ANTHROPIC_AUTH_TOKEN too.

Cursor#

1. Settings → Models → API Keys.
2. Paste your key into OpenAI API Key, switch on Override OpenAI Base URL and enter https://fullsendai.com/v1.
3. Under Models, add a custom model named deepseek-v4.1-flash and select it in chat or the agent.

Cursor's Tab autocomplete always runs on Cursor's own models; chat and agent requests go to FullsendAI. More in the Cursor guide.

Cline and Roo Code#

1. Open the extension's settings and set API Provider to OpenAI Compatible.
2. Base URL: https://fullsendai.com/v1
3. API Key: your fs_live_… key. Model ID: deepseek-v4.1-flash.

Continue#

models:  - name: DeepSeek V4.1 Flash    provider: openai    model: deepseek-v4.1-flash    apiBase: https://fullsendai.com/v1    apiKey: fs_live_...    roles: [chat, edit, apply]

Aider#

export OPENAI_API_BASE=https://fullsendai.com/v1export OPENAI_API_KEY=$FULLSEND_API_KEYaider --model openai/deepseek-v4.1-flash

Codex CLI#

model = "deepseek-v4.1-flash"model_provider = "fullsend"[model_providers.fullsend]name = "FullsendAI"base_url = "https://fullsendai.com/v1"env_key = "FULLSEND_API_KEY"wire_api = "responses"

OpenCode and Kilo Code#

Add FullsendAI as a provider in opencode.json, in your project or in ~/.config/opencode. OpenCode fetches the provider package the first time it runs. Kilo Code is built on OpenCode and reads the same block from ~/.config/kilo/kilo.json, its VS Code extension included.

{  "$schema": "https://opencode.ai/config.json",  "provider": {    "fullsend": {      "npm": "@ai-sdk/openai-compatible",      "name": "FullsendAI",      "options": {        "baseURL": "https://fullsendai.com/v1",        "apiKey": "{env:FULLSEND_API_KEY}"      },      "models": {        "deepseek-v4.1-flash": { "name": "DeepSeek V4.1 Flash", "tool_call": true },        "glm-5.3-flash": { "name": "GLM 5.3 Flash", "tool_call": true }      }    }  },  "model": "fullsend/deepseek-v4.1-flash"}

export FULLSEND_API_KEY=fs_live_...opencode

Anything else#

Most tools have an “OpenAI-compatible” or “custom provider” option. Give it https://fullsendai.com/v1, your key and a model ID. If a tool only lets you change the Anthropic base URL, give it https://fullsendai.com instead. Something not working? Every call, including failed ones, shows in Requests with the reason.