Docs

OpenAI-compatible API documentation

Copy one setup prompt into your existing coding agent, or use the API reference and manual client configurations below.

Paid inference access: start with Qwen3.8-27B using qwen3.8-27b. It is the recommended model for coding, text, and tool use.

Paid alternative: configure yolo if you prefer a stable model ID whose server-side target can change. Existing plan quotas and context caps still apply to every model.

Plan context limits: Free accounts include 0 requests/week and cannot run inference. Legacy Starter allows 128K tokens. Builder, Pro, and Fleet allow 256K tokens. Fleet provides 3x Pro's usage allowances with the same model access and concurrency.

Examples below select qwen3.8-27b. Use GET /v1/models with your API key to discover models before configuring a client. Model discovery does not grant inference access; a paid plan is required. Client context settings below are conservative, not a promise of additional plan capacity.

Install with one prompt

Copy this prompt into the coding agent or AI editor you already use. It will identify the client, ask for your API key, add Yolo-Auto without removing existing providers, and verify the connection.

Configure the coding agent or AI code editor I am currently using to use Yolo-Auto as a model provider. Perform the setup for me instead of only describing it.

Provider details:
- Name: Yolo-Auto
- API protocol: OpenAI-compatible Chat Completions
- Base URL: https://yolo-auto.com/v1
- Default model: qwen3.8-27b
- Context window: 131072
- Use this conservative context setting unless your plan and selected model allow more; model context does not override plan limits.
- API key environment variable: YOLO_AUTO_API_KEY

Instructions:
- Identify the current coding agent or editor, its installed version, and its local user-level configuration before making changes.
- Ask me for my Yolo-Auto API key now if I have not provided it. Prefer a hidden credential prompt when available.
- Before configuring, make an authenticated GET request to https://yolo-auto.com/v1/models. Use qwen3.8-27b only if listed for this key; otherwise use qwen3.8-27b if listed throughout the configuration and instructions below. Stop if neither is available.
- Qwen3.8-27B (qwen3.8-27b) supports text, images, and tools on paid plans. Free accounts have 0 requests/week and must upgrade before sending API requests. yolo is a configurable paid alternative, currently backed by Qwen3.8-27B, whose server-side target may change.
- Treat the API key as a secret. Never print the full key, place it in command history or logs, or store it in a repository.
- Use the agent's native custom OpenAI-compatible provider support when available. Preserve all existing providers and unrelated settings.
- Store the key in the agent's secure credential store or a user-level environment/config file that is excluded from version control.
- Add Yolo-Auto and make qwen3.8-27b available. Do not remove another provider or change my existing default model unless I ask.
- If this agent cannot use an OpenAI-compatible provider directly, use the smallest supported local compatibility bridge. Explain why it is needed and ask before installing a dependency.
- Validate the resulting configuration, then make an authenticated GET request to https://yolo-auto.com/v1/models or use the agent's safest equivalent connection check. Keep the key redacted.
- Finish by reporting the exact files or settings changed and the one command or action I should use to start coding with Yolo-Auto.

Base URL

https://yolo-auto.com/v1

Authentication

Authorization: Bearer yolo_...

Model IDs

"qwen3.8-27b", "yolo", "yolo-small", "yolo-128k"

Prefer an app? Yolo-Auto Desktop is our open-source Windows and Mac client for these models. Download and details · Other releases · GitHub.

OpenAI-Compatible API Endpoints

Use base URL https://yolo-auto.com/v1 and send Authorization: Bearer yolo_YOUR_KEY. Yolo-Auto implements a focused OpenAI-compatible surface rather than every OpenAI API route.

GET /v1/modelsList available public model IDs with the plan-bounded context window and the thinking levels each model accepts.

POST /v1/chat/completionsSend messages and create a normal or streaming chat completion.

GET /v1/usageRead request, token, and quota metadata for the authenticated API key.

Use the request examples below to verify the endpoint, API key, model ID, JSON body, and streaming behavior before configuring a third-party client.

Request format and limits

Qwen3.8-27B (qwen3.8-27b) supports text, images, and tools, as does yolo while backed by the same model. Chat completion request bodies may contain up to 20 MB of JSON.

Existing clients with explicitly configured model IDs remain supported, including their image input capability. Check GET /v1/models with your key for access.

Capacity and retries

Paid plans have fixed monthly prices with no per-token overage fees. Fair use, bounded project queues, and shared-capacity availability apply.

Build clients for transient capacity responses. Requests that exceed your plan allowance return HTTP 429 immediately. Honor Retry-After before retrying. Your subscription price does not change.

When your project is at capacity, requests within the plan allowance can wait in a bounded project queue. A full queue or an expired wait returns HTTP 429.

Admitted requests remain subject to inference capacity and timeouts; admission does not guarantee immediate model execution.

Client guidance

Reduce request volume when requests queue or return HTTP 429.
Cancel abandoned requests and streams promptly.
Honor Retry-After on HTTP 429 responses and retry with jitter.
Do not submit duplicate work while the original request is queued or in progress.

Manual provider references

Use the client-specific sections below if you want to inspect or apply the configuration yourself.

oh-my-pi (omp)Preferred harness

Preferred harness. Registered Yolo-Auto provider plus a plan-first todo-discipline extension.

Yolo-Auto is a registered provider in oh-my-pi. Sign in with your API key once and omp discovers the live model list from GET /v1/models. No provider block, no models.yml entry.

1. Install omp

# macOS / Linux
curl -fsSL https://omp.sh/install | sh

# Windows (PowerShell)
irm https://omp.sh/install.ps1 | iex

2. Register Yolo-Auto

Create a key at yolo-auto.com/app, then attach it to omp. After discovery, select a model with /model or --model.

# Start omp and register the provider (paste your yolo_... key when prompted).
omp
/login yolo-auto

# Or provide the key non-interactively; omp reads it from the environment.
export YOLO_AUTO_API_KEY="yolo_YOUR_KEY_HERE"   # Windows PowerShell: setx YOLO_AUTO_API_KEY "yolo_YOUR_KEY_HERE"

# Confirm discovery of the live model list, then select the model.
omp models yolo-auto
omp --model yolo-auto/qwen3.8-27b

# Optional: make it the session default in ~/.omp/agent/config.yml
modelRoles:
  default: yolo-auto/qwen3.8-27b

3. Recommended: plan-first todo discipline

The yolo-auto/yolo auto-router model performs best with plan-first todo lists. This official extension enforces that ordering: the first non-todo tool call of each request is blocked once with an instruction to commit a plan, and a short per-model addendum rides the system prompt. Other models are untouched; subagent sessions never arm. Full guide: yolo-todo-discipline-setup.md · repo: https://github.com/yolo-auto-org/yolo-auto-omp.

# macOS / Linux
mkdir -p ~/.omp/agent/extensions
curl -fsSL https://raw.githubusercontent.com/yolo-auto-org/yolo-auto-omp/main/yolo-todo-discipline.ts -o ~/.omp/agent/extensions/yolo-todo-discipline.ts

# Windows (PowerShell)
New-Item -ItemType Directory -Force "$env:USERPROFILE\.omp\agent\extensions" | Out-Null
curl.exe -fsSL https://raw.githubusercontent.com/yolo-auto-org/yolo-auto-omp/main/yolo-todo-discipline.ts -o "$env:USERPROFILE\.omp\agent\extensions\yolo-todo-discipline.ts"

# Then start a new omp session (/new or relaunch). Extensions load at session start.

Prompt for your current agent

Paste this into an AI agent that already runs on Yolo-Auto (or another provider) and let it perform the whole setup, extension included.

Configure oh-my-pi (omp) on this machine to use Yolo-Auto. Yolo-Auto is a registered omp provider, so no custom provider block or models.yml entry is needed.

Requirements:
- Ask me for my Yolo-Auto API key if I have not provided it (it starts with yolo_). Never print the full key back to me or write it to a file that is committed to version control.
- Check whether `omp` is installed (`omp --version`). If not, install it: Windows PowerShell: irm https://omp.sh/install.ps1 | iex; macOS/Linux: curl -fsSL https://omp.sh/install | sh. Ask me before installing.
- Register the key: prefer the interactive in-session flow (/login yolo-auto inside omp); otherwise set the YOLO_AUTO_API_KEY environment variable for my user shell profile.
- Validate the key before finishing: an authenticated GET to https://yolo-auto.com/v1/models must list models, and `omp models yolo-auto` should discover them. Use qwen3.8-27b in the steps below if it is listed; otherwise use whichever model the listing offers.
- Set the default model role by adding to ~/.omp/agent/config.yml (on Windows: %USERPROFILE%\.omp\agent\config.yml), preserving existing keys:
```yaml
modelRoles:
default: yolo-auto/qwen3.8-27b
```
- Install the recommended Yolo-Auto todo-discipline extension so the auto-router model commits a plan before working. Download it into the omp user extensions directory (~/.omp/agent/extensions/, or %USERPROFILE%\.omp\agent\extensions\ on Windows):
https://raw.githubusercontent.com/yolo-auto-org/yolo-auto-omp/main/yolo-todo-discipline.ts
Extensions load at session start, so a new omp session is required after installing.
- Do not remove or alter other providers or unrelated omp settings.
- Report the exact files changed and the one command to start coding with Yolo-Auto.

Pi

Configure Pi models.json.

Use this provider block in your Pi model configuration.

{
  "providers": {
    "Yolo-Auto": {
      "baseUrl": "https://yolo-auto.com/v1",
      "api": "openai-completions",
      "apiKey": "YOUR_yolo_API_KEY_HERE",
      "compat": {
        "supportsDeveloperRole": false,
        "supportsStore": false,
        "supportsReasoningEffort": true,
        "thinkingFormat": "qwen-chat-template"
      },
      "models": [
        {
          "id": "qwen3.8-27b",
          "name": "Qwen3.8-27B",
          "reasoning": true,
          "input": [
            "text",
            "image"
          ],
          "contextWindow": 131072,
          "cost": {
            "input": 0,
            "output": 0,
            "cacheRead": 0,
            "cacheWrite": 0
          }
        }
      ]
    }
  }
}

Prompt for your current agent

Paste this into your current coding agent and let it make the Pi config changes.

Configure this machine's Pi agent to use Yolo-Auto.

Requirements:
- Ask me for my Yolo-Auto API key if I have not provided it, and do not print the full key back to me.
- Verify https://yolo-auto.com/v1/models with my key before setup. If qwen3.8-27b is unavailable, use qwen3.8-27b if listed instead throughout the config and instructions below. Stop if neither is available.
- Preserve all existing Pi providers, models, auth, comments/format as much as possible.
- Update or create ~/.pi/agent/models.json and merge in this provider block under providers.Yolo-Auto:
{
  "providers": {
    "Yolo-Auto": {
      "baseUrl": "https://yolo-auto.com/v1",
      "api": "openai-completions",
      "apiKey": "PASTE_YOUR_YOLO_AUTO_API_KEY_HERE",
      "compat": {
        "supportsDeveloperRole": false,
        "supportsStore": false,
        "supportsReasoningEffort": true,
        "thinkingFormat": "qwen-chat-template"
      },
      "models": [
        {
          "id": "qwen3.8-27b",
          "name": "Qwen3.8-27B",
          "reasoning": true,
          "input": [
            "text",
            "image"
          ],
          "contextWindow": 131072,
          "cost": {
            "input": 0,
            "output": 0,
            "cacheRead": 0,
            "cacheWrite": 0
          }
        }
      ]
    }
  }
}
- If Pi has a default model/provider setting, set the default model to qwen3.8-27b for Yolo-Auto only when that setting exists and can be changed safely.
- Validate the JSON after editing.
- Report the exact files changed and a one-command way for me to test it.

OpenClaw

Configure an OpenClaw agent models.json.

OpenClaw stores agent config under ~/.openclaw/agents/. For the agent you run, update agents/<agent-name>/models.json.

{
  "providers": {
    "yolo-auto": {
      "baseUrl": "https://yolo-auto.com/v1",
      "apiKey": "ENTER_YOUR_yolo_KEY_HERE",
      "api": "openai-completions",
      "models": [
        {
          "id": "qwen3.8-27b",
          "name": "Qwen3.8-27B",
          "contextWindow": 128000,
          "reasoning": true,
          "input": [
            "text",
            "image"
          ]
        }
      ]
    }
  }
}

Prompt for your current agent

Paste this into your current coding agent and let it update the selected OpenClaw agent.

Configure my existing OpenClaw agent to use Yolo-Auto.

Requirements:
- Ask me which OpenClaw agent name to configure if it is not obvious from ~/.openclaw/agents/.
- Ask me for my Yolo-Auto API key if I have not provided it, and do not print the full key back to me.
- Verify https://yolo-auto.com/v1/models with my key before setup. If qwen3.8-27b is unavailable, use qwen3.8-27b if listed instead throughout the config and instructions below. Stop if neither is available.
- Preserve all existing providers and unrelated settings.
- Update or create ~/.openclaw/agents/<agent-name>/models.json and merge in this provider block under providers.yolo-auto:
{
  "providers": {
    "yolo-auto": {
      "baseUrl": "https://yolo-auto.com/v1",
      "apiKey": "PASTE_YOUR_YOLO_AUTO_API_KEY_HERE",
      "api": "openai-completions",
      "models": [
        {
          "id": "qwen3.8-27b",
          "name": "Qwen3.8-27B",
          "contextWindow": 128000,
          "reasoning": true,
          "input": [
            "text",
            "image"
          ]
        }
      ]
    }
  }
}
- Keep the model id qwen3.8-27b and contextWindow 128000.
- Validate the JSON after editing.
- Report the exact files changed and how I should restart/test OpenClaw.

Hermes Agent

Set Hermes custom endpoint values.

Follow Hermes Agent's custom endpoint setup and use these values.

Provider: custom
Base URL: https://yolo-auto.com/v1
API key: YOUR_yolo_API_KEY_HERE
Default model: qwen3.8-27b
API mode: chat_completions
Context length: 131072

Prompt for your current agent

Paste this into your current coding agent and let it configure Hermes.

Configure Hermes Agent to use Yolo-Auto as its custom OpenAI-compatible endpoint.

Requirements:
- Ask me for my Yolo-Auto API key if I have not provided it, and do not print the full key back to me.
- Verify https://yolo-auto.com/v1/models with my key before setup. If qwen3.8-27b is unavailable, use qwen3.8-27b if listed instead throughout the config and instructions below. Stop if neither is available.
- Preserve existing Hermes config except the main model provider/base/model values needed below.
- Set these values:
Provider: custom
Base URL: https://yolo-auto.com/v1
API key: PASTE_YOUR_YOLO_AUTO_API_KEY_HERE
Default model: qwen3.8-27b
API mode: chat_completions
Context length: 131072
- Prefer Hermes' CLI when available:
  hermes config set model.provider custom
  hermes config set model.base_url https://yolo-auto.com/v1
  hermes config set model.default qwen3.8-27b
  hermes config set model.context_length 131072
  hermes config set model.api_mode chat_completions
- Also set OPENAI_API_KEY and OPENAI_BASE_URL in the Hermes .env location for this OS/HERMES_HOME if Hermes uses it.
- Validate by showing the sanitized effective provider/base/model settings, with the API key redacted.
- Report the exact files/commands changed and how I should restart/test Hermes.

OpenCode

Add a Yolo-Auto OpenAI-compatible provider.

Add this provider to opencode.json. Keep your real key in YOLO_AUTO_API_KEY.

{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "yolo-auto": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Yolo-Auto",
      "options": {
        "baseURL": "https://yolo-auto.com/v1",
        "apiKey": "{env:YOLO_AUTO_API_KEY}"
      },
      "models": {
        "qwen3.8-27b": {
          "name": "Qwen3.8-27B",
          "limit": {
            "context": 131072,
            "output": 32768
          },
          "attachment": true,
          "modalities": {
            "input": [
              "text",
              "image"
            ],
            "output": [
              "text"
            ]
          }
        }
      }
    }
  }
}

Prompt for your current agent

Paste this into your current coding agent and let it configure OpenCode.

Configure OpenCode to use Yolo-Auto.

Requirements:
- Ask me for my Yolo-Auto API key if I have not provided it, and do not print the full key back to me.
- Verify https://yolo-auto.com/v1/models with my key before setup. If qwen3.8-27b is unavailable, use qwen3.8-27b if listed instead throughout the config and instructions below. Stop if neither is available.
- Preserve existing OpenCode providers and unrelated settings.
- Update or create the appropriate opencode.json for this workspace/user.
- Add or merge this provider under provider.yolo-auto:
{
  "$schema": "https://opencode.ai/config.json",
  "provider": {
    "yolo-auto": {
      "npm": "@ai-sdk/openai-compatible",
      "name": "Yolo-Auto",
      "options": {
        "baseURL": "https://yolo-auto.com/v1",
        "apiKey": "{env:YOLO_AUTO_API_KEY}"
      },
      "models": {
        "qwen3.8-27b": {
          "name": "Qwen3.8-27B",
          "limit": {
            "context": 131072,
            "output": 32768
          },
          "attachment": true,
          "modalities": {
            "input": [
              "text",
              "image"
            ],
            "output": [
              "text"
            ]
          }
        }
      }
    }
  }
}
- Store the real key in YOLO_AUTO_API_KEY or OpenCode's normal credential store; do not hardcode the secret into a committed repo file unless I explicitly ask.
- Configure/select model qwen3.8-27b.
- Validate the JSON after editing.
- Report the exact files changed and how I should run/test OpenCode.

Aider

Use Aider with an OpenAI-compatible base URL.

Aider uses an OpenAI-compatible base URL and the openai/ model prefix.

export AIDER_OPENAI_API_BASE="https://yolo-auto.com/v1"
export AIDER_OPENAI_API_KEY="YOUR_yolo_API_KEY_HERE"
export AIDER_MODEL="openai/qwen3.8-27b"

# Or run one-off: aider --openai-api-base "https://yolo-auto.com/v1" --openai-api-key "YOUR_yolo_API_KEY_HERE" --model openai/qwen3.8-27b

Prompt for your current agent

Paste this into your current coding agent and let it configure Aider.

Configure Aider to use Yolo-Auto through its OpenAI-compatible settings.

Requirements:
- Ask me for my Yolo-Auto API key if I have not provided it, and do not print the full key back to me.
- Verify https://yolo-auto.com/v1/models with my key before setup. If qwen3.8-27b is unavailable, use qwen3.8-27b if listed instead throughout the config and instructions below. Stop if neither is available.
- Prefer project-local config/env files only if they are gitignored; otherwise tell me what shell exports to add.
- Set these values:
export AIDER_OPENAI_API_BASE="https://yolo-auto.com/v1"
export AIDER_OPENAI_API_KEY="PASTE_YOUR_YOLO_AUTO_API_KEY_HERE"
export AIDER_MODEL="openai/qwen3.8-27b"

# Or run one-off: aider --openai-api-base "https://yolo-auto.com/v1" --openai-api-key "PASTE_YOUR_YOLO_AUTO_API_KEY_HERE" --model openai/qwen3.8-27b
- Use model openai/qwen3.8-27b.
- Do not commit or expose the API key.
- Report the exact files changed and a one-command test using aider.

Cline

Use the OpenAI Compatible provider in VS Code.

In Cline settings, choose OpenAI Compatible, then enter these values.

API Provider: OpenAI Compatible
Base URL: https://yolo-auto.com/v1
API Key: YOUR_yolo_API_KEY_HERE
Model ID: qwen3.8-27b
Context window: 131072
Supports images: true

Prompt for your current agent

Paste this into your current coding agent and let it configure Cline or give exact UI steps.

Configure Cline to use Yolo-Auto via the OpenAI Compatible provider.

Requirements:
- Ask me for my Yolo-Auto API key if I have not provided it, and do not print the full key back to me.
- Verify https://yolo-auto.com/v1/models with my key before setup. If qwen3.8-27b is unavailable, use qwen3.8-27b if listed instead throughout the config and instructions below. Stop if neither is available.
- If Cline settings can be safely edited as JSON on this machine, update them; otherwise give me exact UI steps.
- Use these values:
API Provider: OpenAI Compatible
Base URL: https://yolo-auto.com/v1
API Key: PASTE_YOUR_YOLO_AUTO_API_KEY_HERE
Model ID: qwen3.8-27b
Context window: 131072
Supports images: true
- Preserve unrelated Cline/VS Code settings.
- Report the exact files changed or the exact UI fields I need to set.

Roo Code

Use Roo Code's OpenAI Compatible provider.

In Roo Code settings, choose OpenAI Compatible, then enter these values.

API Provider: OpenAI Compatible
Base URL: https://yolo-auto.com/v1
API Key: YOUR_yolo_API_KEY_HERE
Model ID: qwen3.8-27b
Context window: 131072
Note: Roo Code requires OpenAI-compatible native tool calling from the selected model.

Prompt for your current agent

Paste this into your current coding agent and let it configure Roo Code or give exact UI steps.

Configure Roo Code to use Yolo-Auto via the OpenAI Compatible provider.

Requirements:
- Ask me for my Yolo-Auto API key if I have not provided it, and do not print the full key back to me.
- Verify https://yolo-auto.com/v1/models with my key before setup. If qwen3.8-27b is unavailable, use qwen3.8-27b if listed instead throughout the config and instructions below. Stop if neither is available.
- If Roo Code settings can be safely edited as JSON on this machine, update them; otherwise give me exact UI steps.
- Use these values:
API Provider: OpenAI Compatible
Base URL: https://yolo-auto.com/v1
API Key: PASTE_YOUR_YOLO_AUTO_API_KEY_HERE
Model ID: qwen3.8-27b
Context window: 131072
Note: Roo Code requires OpenAI-compatible native tool calling from the selected model.
- Preserve unrelated Roo Code/VS Code settings.
- Warn me if the selected model/tool-calling compatibility cannot be verified locally.
- Report the exact files changed or the exact UI fields I need to set.

Continue

Add a Yolo-Auto model to Continue config.

Add this model entry to Continue's config YAML.

name: Yolo-Auto
version: 0.0.1
schema: v1

models:
  - name: Yolo-Auto Qwen3.8-27B
    provider: openai
    model: qwen3.8-27b
    apiBase: https://yolo-auto.com/v1
    apiKey: YOUR_yolo_API_KEY_HERE
    useResponsesApi: false

Prompt for your current agent

Paste this into your current coding agent and let it configure Continue.

Configure Continue to use Yolo-Auto as an OpenAI-compatible model provider.

Requirements:
- Ask me for my Yolo-Auto API key if I have not provided it, and do not print the full key back to me.
- Verify https://yolo-auto.com/v1/models with my key before setup. If qwen3.8-27b is unavailable, use qwen3.8-27b if listed instead throughout the config and instructions below. Stop if neither is available.
- Locate Continue's config YAML/JSON for this install and preserve unrelated models/settings.
- Add or merge this model entry:
name: Yolo-Auto
version: 0.0.1
schema: v1

models:
  - name: Yolo-Auto Qwen3.8-27B
    provider: openai
    model: qwen3.8-27b
    apiBase: https://yolo-auto.com/v1
    apiKey: PASTE_YOUR_YOLO_AUTO_API_KEY_HERE
    useResponsesApi: false
- Keep useResponsesApi: false so Continue uses chat completions.
- Do not commit or expose the API key.
- Validate the config syntax and report how I should restart/test Continue.

Claude Code

Bridge through LiteLLM's Anthropic-compatible endpoint.

Claude Code expects Anthropic Messages API. Use LiteLLM as a local bridge to Yolo-Auto's OpenAI-compatible endpoint.

1. Save this as `litellm-yolo-auto.yaml`

model_list:
  - model_name: qwen3.8-27b
    litellm_params:
      model: openai/qwen3.8-27b
      api_key: os.environ/YOLO_AUTO_API_KEY
      api_base: https://yolo-auto.com/v1

litellm_settings:
  master_key: os.environ/LITELLM_MASTER_KEY

2. Start the bridge and launch Claude Code

# Install LiteLLM proxy support
uv tool install 'litellm[proxy]'

# Set secrets locally. Use your real yolo_ key and generate your own proxy key.
export YOLO_AUTO_API_KEY="YOUR_yolo_API_KEY_HERE"
export LITELLM_MASTER_KEY="sk-local-proxy-key-change-me"

# Start the Anthropic-compatible bridge. Save the YAML above as litellm-yolo-auto.yaml first.
litellm --config ./litellm-yolo-auto.yaml --host 127.0.0.1 --port 4000

# In another shell, point Claude Code at LiteLLM's Anthropic Messages API bridge.
export ANTHROPIC_BASE_URL="http://127.0.0.1:4000"
export ANTHROPIC_AUTH_TOKEN="$LITELLM_MASTER_KEY"
export CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY=1
claude --model qwen3.8-27b

Prompt for your current agent

Paste this into your current coding agent and let it set up the LiteLLM bridge.

Configure Claude Code to use Yolo-Auto through a local LiteLLM bridge.

Context:
- Claude Code expects an Anthropic Messages API endpoint.
- Yolo-Auto is OpenAI-compatible, so LiteLLM should bridge Claude Code -> LiteLLM /v1/messages -> Yolo-Auto /v1/chat/completions.

Requirements:
- Ask me for my Yolo-Auto API key if I have not provided it, and do not print the full key back to me.
- Verify https://yolo-auto.com/v1/models with my key before setup. If qwen3.8-27b is unavailable, use qwen3.8-27b if listed instead throughout the config and instructions below. Stop if neither is available.
- Install LiteLLM proxy support only if needed and after confirming the package manager/uv availability.
- Create or update a local litellm-yolo-auto.yaml with:
model_list:
  - model_name: qwen3.8-27b
    litellm_params:
      model: openai/qwen3.8-27b
      api_key: os.environ/YOLO_AUTO_API_KEY
      api_base: https://yolo-auto.com/v1

litellm_settings:
  master_key: os.environ/LITELLM_MASTER_KEY
- Set local environment variables without committing secrets:
  YOLO_AUTO_API_KEY=<my yolo_ key>
  LITELLM_MASTER_KEY=<a generated local proxy key>
  ANTHROPIC_BASE_URL=http://127.0.0.1:4000
  ANTHROPIC_AUTH_TOKEN=$LITELLM_MASTER_KEY
  CLAUDE_CODE_ENABLE_GATEWAY_MODEL_DISCOVERY=1
- Show me the command to start LiteLLM and the command to launch Claude Code with qwen3.8-27b.
- Report the exact files changed and keep all secrets redacted.

Request examples

List models

Each entry reports context_length and max_model_len for the window your plan allows on that model, and thinking with the accepted reasoning levels when the model supports them. Configure client context windows from this response instead of a bundled model catalog, which can overstate the window your plan enforces.

curl https://yolo-auto.com/v1/models \
  -H "Authorization: Bearer yolo_YOUR_KEY"

Chat completion

Streaming is recommended for long-context or long-output requests. When the upstream starts the response promptly, streaming reduces the risk of hitting the proxied read deadline before inference finishes.

curl https://yolo-auto.com/v1/chat/completions \
  -H "Authorization: Bearer yolo_YOUR_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "qwen3.8-27b",
    "messages": [
      { "role": "user", "content": "Hello from Yolo-Auto" }
    ],
    "stream": true
  }'

Usage

curl https://yolo-auto.com/v1/usage \
  -H "Authorization: Bearer yolo_YOUR_KEY"

OpenAI SDK

import OpenAI from "openai";

const client = new OpenAI({
  apiKey: process.env.YOLO_AUTO_API_KEY,
  baseURL: "https://yolo-auto.com/v1"
});

const response = await client.chat.completions.create({
  model: "qwen3.8-27b",
  messages: [{ role: "user", content: "Hello" }]
});

console.log(response.choices[0]?.message?.content);

OpenAI-compatible API documentation

Install with one prompt

Base URL

Authentication

Model IDs

OpenAI-Compatible API Endpoints

Request format and limits

Capacity and retries

Client guidance

Manual provider references

oh-my-pi (omp)Preferred harness

1. Install omp

2. Register Yolo-Auto

3. Recommended: plan-first todo discipline

Prompt for your current agent

Pi

Prompt for your current agent

OpenClaw

Prompt for your current agent

Hermes Agent

Prompt for your current agent

OpenCode

Prompt for your current agent

Aider

Prompt for your current agent

Cline

Prompt for your current agent

Roo Code

Prompt for your current agent

Continue

Prompt for your current agent

Claude Code

1. Save this as litellm-yolo-auto.yaml

2. Start the bridge and launch Claude Code

Prompt for your current agent

Request examples

List models

Chat completion

Usage

OpenAI SDK

1. Save this as `litellm-yolo-auto.yaml`