Clients/All clients

Client Setup

Every client needs only the Wokey base URL and an API key. Pick your client to see its configuration.

On this pageexpand_more

Terminal coding agents
Desktop and chat clients
Always-on agents
Model gateways
Base URL reference

Terminal coding agentslink

Agents that read and write code and run commands in a terminal, configured through environment variables or a config file.

Claude Codearrow_forwardClaude Code uses Anthropic-compatible environment variables and sends ANTHROPIC_AUTH_TOKEN as an Authorization Bearer token.Default format: Anthropic Messages Codexarrow_forwardCodex CLI and the Codex desktop app share the same configuration, so the setup below applies to both. Codex uses the Responses protocol. Keep base_url at the gateway origin, without /responses. Replace YOUR_API_KEY with your Wokey API key in experimental_bearer_token, without a Bearer prefix. Remove any existing env_key setting; no environment variable export is needed. Save the file and run codex. The key is stored in plain text in the local config.Default format: OpenAI Responses opencodearrow_forwardOne command writes the Wokey provider and switches the default model to wokey/auto.Default format: OpenAI Chat Completions Grok CLIarrow_forwardUse the official Grok CLI config to connect to Wokey. Add a grok-4.7 model entry with its own api_key; the api_key in a model entry takes precedence over the grok login session, so Responses requests go to the Wokey gateway.Default format: OpenAI Responses piarrow_forwardpi reads custom providers from models.json. The configuration below calls Claude models through Anthropic Messages; for GPT, DeepSeek, and other models add a second provider with api set to openai-completions and a baseUrl ending in /v1.Default format: Anthropic Messages DeepSeek Harnessarrow_forwardDeepSeek Harness (dsh) accepts custom gateways, so one Wokey key covers DeepSeek, GPT, GLM and Claude models. deepseek-v4.1-flash supports Chat Completions, Responses, and Anthropic Messages; the configuration below puts Claude Opus 5.5 and DeepSeek-V4.1-Flash under one Anthropic Messages provider and declares image input for both.Default format: OpenAI Chat Completions

Desktop and chat clientslink

Apps with a graphical interface, where you add a custom provider on the settings page.

Claude Desktoparrow_forwardClaude Desktop requires the latest app and uses developer mode to configure third-party inference. Follow the steps on the left.Default format: Anthropic Messages Z Codearrow_forwardZ Code adds custom providers on its settings page and supports both the Anthropic and OpenAI API formats. Use the Anthropic format for Claude models.Default format: Anthropic Messages Cherry Studioarrow_forwardCherry Studio adds custom providers under Model Provider. The OpenAI endpoint is enough to call every Wokey model through Chat Completions, and the same provider can also take an Anthropic endpoint.Default format: OpenAI Chat Completions WorkBuddyarrow_forwardWorkBuddy uses OpenAI Chat Completions for custom models. In WorkBuddy, open Settings → Model → Custom models and enter Wokey’s endpoint, API key, and deepseek-v4.1-flash; tool calls and image input can then be used by WorkBuddy tasks and agent workflows.Default format: OpenAI Chat Completions

Always-on agentslink

Agents that keep running on your own machine or server and take tasks through chat apps.

OpenClawarrow_forwardWrite a custom OpenAI-compatible provider with one command.Default format: OpenAI Chat Completions Hermes Agentarrow_forwardHermes Agent only needs the first five config lines for model, custom provider, Base URL, and API key. For GPT models (such as gpt-6-sol), add api_mode: codex_responses for the best cache hit rate; leave it off for other models such as Claude.Default format: OpenAI Chat Completions

Model gatewayslink

Gateways that run on your machine, use Wokey as an upstream, and serve several agents from it.

Magpiearrow_forwardMagpie is a model gateway that runs on your machine. Once Wokey is added as a provider, every agent on the machine can call Wokey models through it. Magpie stores one address per API format, and a single command sets all three.Default format: OpenAI Chat Completions

Base URL referencelink

The base URL each client takes for each API format, and where the client keeps the setting. Clients without a page of their own are listed too.

Client	Format	Base URL	Where it is configured
Claude Code	Anthropic Messages	https://api.wokey.ai	Environment variables, or env in ~/.claude/settings.json
Codex	OpenAI Responses	https://api.wokey.ai	~/.codex/config.toml
Claude Desktop	Anthropic Messages	https://api.wokey.ai	Menu Developer → Configure third-party inference
opencode	OpenAI Chat Completions	https://api.wokey.ai	~/.config/opencode/opencode.json
Grok CLI	OpenAI Responses	https://api.wokey.ai	~/.grok/config.toml
pi	Anthropic MessagesOpenAI Chat CompletionsOpenAI Responses	https://api.wokey.aihttps://api.wokey.ai/v1https://api.wokey.ai/v1	~/.pi/agent/models.json
DeepSeek Harness	OpenAI Chat CompletionsAnthropic Messages	https://api.wokey.ai/v1https://api.wokey.ai	~/.dsh/profiles/web/cordis.patch.yml
Z Code	Anthropic MessagesOpenAI Chat Completions	https://api.wokey.aihttps://api.wokey.ai/v1	Settings → Model Settings (file ~/.zcode/v2/config.json)
Cherry Studio	OpenAI Chat CompletionsAnthropic Messages	https://api.wokey.aihttps://api.wokey.ai	Settings → Model Provider → Add Provider
WorkBuddy	OpenAI Chat Completions	https://api.wokey.ai	Settings → Models → Custom models
OpenClaw	OpenAI Chat Completions	https://api.wokey.ai	openclaw onboard command flags
Hermes Agent	OpenAI Chat Completions	https://api.wokey.ai	~/.hermes/config.yaml
Magpie	OpenAI Chat CompletionsOpenAI ResponsesAnthropic Messages	https://api.wokey.ai/v1https://api.wokey.ai/v1https://api.wokey.ai	magpie provider add or an import link (file ~/.config/magpie/providers.json)
Cline / Kilo Code / Roo Code	OpenAI Chat Completions	https://api.wokey.ai/v1	Extension settings, choose OpenAI Compatible (Kilo Code: Providers → Custom provider)
Aider	OpenAI Chat Completions	https://api.wokey.ai/v1	Environment variables OPENAI_API_BASE and OPENAI_API_KEY; model as openai/model-name
Zed	OpenAI Chat Completions	https://api.wokey.ai/v1	language_models.openai_compatible in settings.json
n8n	OpenAI Chat Completions	https://api.wokey.ai/v1	Base URL in the OpenAI credential

Calling the API directly with curl or an SDK? Request examples for each format are linked from Endpoints