Connect RouterBrain in any AI tool that accepts a custom OpenAI- or Anthropic-compatible endpoint.

Dedicated guides exist for Cursor, Claude Code, OpenClaw, TRAE, Hermes Agent, and Codex. If your tool supports Base URL + API Key but is not listed below, use the reference values.

Create an inference API key in the console (API keys) and pick a catalog id (List models) first.

Configuration reference

RouterBrain exposes two compatible surfaces—use whichever your tool supports.

OpenAI-compatible

FieldValue
ProviderOpenAI Compatible (sometimes Custom or OpenAI-format)
Base URLhttps://51kik.com/v1
API KeyConsole inference key
Model IDCatalog id

Anthropic-compatible

FieldValue
ProviderAnthropic Compatible (sometimes Claude or Custom Anthropic)
Base URLhttps://51kik.com/anthropic
API KeyConsole inference key
Model IDCatalog id

Which protocol to pick

Tool styleRecommendedCommon env vars
Claude Code–style Anthropic TUI/CLIAnthropic-compatibleANTHROPIC_BASE_URL + ANTHROPIC_AUTH_TOKEN
Cursor / Continue / Aider / OpenAI IDE pluginsOpenAI-compatibleOPENAI_BASE_URL + OPENAI_API_KEY
Both supportedEither; Anthropic surface → Messages API

Minimal check

curl -sS "https://51kik.com/v1/chat/completions" \
  -H "Authorization: Bearer YOUR_ROUTERBRAIN_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model":"YOUR_MODEL_ID","messages":[{"role":"user","content":"ok"}]}'

Dify

Dify is an open-source LLM app platform with workflows, RAG, agents, and observability.

  1. Open Dify Cloud or your own Dify deployment → Settings → Workspace → Model providers.
  2. Install OpenAI-API-compatible (or Anthropic-compatible, depending on version).
  3. Set API Key (console key) and API Base (OpenAI: https://51kik.com/v1; Anthropic: https://51kik.com/anthropic).
  4. Add your catalog id as the model name and use it in workflows.

Cherry Studio

Cherry Studio is an open-source desktop client with MCP support.

  1. Install per official docs.
  2. Choose other ProvidersOpenAI Compatible (or custom OpenAI protocol).
  3. API Host / Base URLhttps://51kik.com/v1, API Key → console key, Check to verify.
  4. Add a custom model name (catalog id) and use it.

Xcode

Xcode includes Coding Intelligence with Internet Hosted providers.

  1. Install Xcode 26 or newer.
  2. Xcode → Settings → Intelligence → Add a Model Provider → Internet Hosted:
    • URL: https://51kik.com (gateway root; if the UI requires a path, try https://51kik.com/v1 per Apple docs)
    • API Key Header: Authorization
    • API Key: Bearer YOUR_ROUTERBRAIN_KEY (one space after Bearer)
    • Description: RouterBrain (any label)
  3. Enable the model; ⌘+0 for Coding Assistant and select it.

Xcode third-party fields vary by version—validate key and model with curl if connection fails.


Pi

Pi is an open-source terminal coding agent.

npm install -g @earendil-works/pi-coding-agent

export OPENAI_API_KEY=YOUR_ROUTERBRAIN_KEY
export OPENAI_BASE_URL=https://51kik.com/v1
pi --provider openai --model YOUR_MODEL_ID

Custom provider names depend on your Pi version—check pi --help.


Kilo Code

Kilo Code is a VS Code AI agent extension.

Unset ANTHROPIC_AUTH_TOKEN and ANTHROPIC_BASE_URL first so they do not override plugin settings.

  1. Install Kilo Code from the marketplace.
  2. Settings:
    • API ProviderOpenAI Compatible / Custom
    • Base URLhttps://51kik.com/v1
    • API Key → console key
    • Model → catalog id
  3. Save and use.

For Anthropic-style tools, use https://51kik.com/anthropic per Kilo Code docs.


Zed

Zed is a high-performance collaborative editor.

  1. Install via Zed docs.
  2. Settings → LLM Provider → +Add Provider → OpenAI:
    • API URL: https://51kik.com/v1
    • API Key: console key
    • Model Name: catalog id
  3. Save Provider; some builds ask you to confirm the key again.
  4. Select a Model in the agent panel.

OpenCode

OpenCode is SST's terminal coding agent.

curl -fsSL https://opencode.ai/install | bash
# or npm i -g opencode-ai

Configure a custom OpenAI-compatible endpoint (path varies by version—~/.opencode/config or Custom during opencode auth login):

  • Base URL: https://51kik.com/v1
  • API Key: console key
  • Model: YOUR_MODEL_ID

Grok CLI

Grok CLI accepts OpenAI-compatible backends.

For heavy agent workflows, prefer Claude Code or Cursor.

Unset OPENAI_API_KEY and OPENAI_BASE_URL first.

npm install -g @vibe-kit/grok-cli

export GROK_BASE_URL=https://51kik.com/v1
export GROK_API_KEY=YOUR_ROUTERBRAIN_KEY
grok --model YOUR_MODEL_ID

Droid

Droid is Factory's terminal coding agent.

Unset ANTHROPIC_AUTH_TOKEN. Config file: ~/.factory/config.json (not settings.json).

curl -fsSL https://app.factory.ai/cli | sh

Edit ~/.factory/config.json:

{
  "custom_models": [{
    "model_display_name": "RouterBrain",
    "model": "YOUR_MODEL_ID",
    "base_url": "https://51kik.com/anthropic",
    "api_key": "YOUR_ROUTERBRAIN_KEY",
    "provider": "anthropic",
    "max_tokens": 64000
  }]
}

Run droid, /model to select the model.


MonkeyCode

MonkeyCode is Chaitin's enterprise AI dev platform.

  1. Sign in at monkeycode-ai.com.
  2. Settings → AI models → Bind.
  3. Set:
    • API URL: https://51kik.com/anthropic (Anthropic) or https://51kik.com/v1 (OpenAI)
    • API Key: console key
    • Format: anthropic or openai
    • Model name: catalog id
  4. Save and use.

Qwen Code

Qwen Code is Alibaba's terminal coding agent.

bash -c "$(curl -fsSL https://qwen-code-assets.oss-cn-hangzhou.aliyuncs.com/installation/install-qwen.sh)"

Run qwenThird-party ProvidersOpenAI Compatible:

  • Base URL: https://51kik.com/v1
  • API Key: console key
  • Model ID: catalog id

Open WebUI

Open WebUI is an open-source chat UI you can run locally.

pip install open-webui
open-webui serve
  1. Open http://localhost:8080 and create an admin account.
  2. Admin Panel → Settings → Connections → OpenAI → Add Connection:
    • URL: https://51kik.com/v1
    • Auth: console key (Bearer)
  3. Pick model id in chat.

nanobot

nanobot is a lightweight personal agent CLI.

uv tool install nanobot-ai
nanobot onboard

Edit ~/.nanobot/config.json (field names may vary by version):

{
  "providers": {
    "openai": {
      "apiKey": "YOUR_ROUTERBRAIN_KEY",
      "apiBase": "https://51kik.com/v1"
    }
  },
  "agents": {
    "defaults": {
      "provider": "openai",
      "model": "YOUR_MODEL_ID"
    }
  }
}
nanobot agent

OpenHands

OpenHands (formerly OpenDevin) is an open-source coding agent.

pipx install openhands

Example setup (adjust to your OpenHands version):

pipx run --spec openhands python -c '
import sys
from openhands_cli.stores.agent_store import AgentStore
AgentStore().create_and_save_from_settings(
    llm_api_key=sys.argv[1],
    settings={
        "llm_model": "openai/YOUR_MODEL_ID",
        "llm_base_url": "https://51kik.com/v1",
    },
)
' YOUR_ROUTERBRAIN_KEY
openhands

LangChain

LangChain is an LLM application framework.

pip install langchain-openai
from langchain_openai import ChatOpenAI

llm = ChatOpenAI(
    model="YOUR_MODEL_ID",
    api_key="YOUR_ROUTERBRAIN_KEY",
    base_url="https://51kik.com/v1",
)
print(llm.invoke("Hello").content)

For Anthropic, use langchain-anthropic with base_url="https://51kik.com/anthropic/v1" (From Anthropic).


Related