Skip to content
NewDeepSeek V4.1 Flash 1017 Uncensored is live, with 1M tokens of context.Read the launch
Coding agents

Other tools

Connect Cline, Roo Code, Continue, Aider, Zed, Hermes Agent and any other OpenAI-compatible client to XField.

Most coding tools that accept a custom OpenAI base URL work with XField unchanged. The recipe is always the same: base URL, key, exact model id. This page covers the common ones and ends with a checklist for everything else.

Every example assumes your key is in XFIELD_API_KEY. Get one from Settings › API keys, or request access.

Cline and Roo Code

Both have an OpenAI Compatible provider in their settings panel.

FieldValue
API ProviderOpenAI Compatible
Base URLhttps://xfield.cloud.brutal.ai/api/v1
API Keyyour key
Model IDdeepseek-v4.1-flash-1017-uncensored

Under the model configuration, set the context window to match the model (1,000,000 for DeepSeek and GLM, 131,072 for Qwen) so the extension knows when to condense. Turn off image support for every model except qwen-3.8-27b-uncensored.

Continue

Add a model to ~/.continue/config.yaml with the openai provider and a custom apiBase:

~/.continue/config.yaml
YAML
name: My config
version: 0.0.1
schema: v1
models:
  - name: DeepSeek V4.1 Flash
    provider: openai
    model: deepseek-v4.1-flash-1017-uncensored
    apiBase: https://xfield.cloud.brutal.ai/api/v1
    apiKey: ${{ secrets.XFIELD_API_KEY }}
    roles: [chat, edit, apply]
  - name: GLM 5.3 Flash
    provider: openai
    model: glm-5.3-flash-uncensored
    apiBase: https://xfield.cloud.brutal.ai/api/v1
    apiKey: ${{ secrets.XFIELD_API_KEY }}
    roles: [summarize]

Local secrets come from a .env file in ~/.continue/. You can also paste the key directly, but then keep the file out of version control.

Aider

Aider reaches OpenAI-compatible APIs through two environment variables and an openai/ prefix on the model name:

Shell
export OPENAI_API_BASE="https://xfield.cloud.brutal.ai/api/v1"
export OPENAI_API_KEY="$XFIELD_API_KEY"
 
aider --model openai/deepseek-v4.1-flash-1017-uncensored

Aider warns that it does not know the model's context size or pricing. That is expected; add --no-show-model-warnings to silence it. Use openai/glm-5.3-uncensored as the --model for hard problems and openai/glm-5.3-flash-uncensored as the --weak-model for commit messages and summaries.

Zed

Add XField under language_models.openai_compatible in Zed's settings.json:

settings.json
JSON
{
  "language_models": {
    "openai_compatible": {
      "XField": {
        "api_url": "https://xfield.cloud.brutal.ai/api/v1",
        "available_models": [
          {
            "name": "deepseek-v4.1-flash-1017-uncensored",
            "display_name": "DeepSeek V4.1 Flash",
            "max_tokens": 1000000,
            "max_output_tokens": 65536,
            "capabilities": {
              "tools": true,
              "images": false,
              "parallel_tool_calls": false,
              "prompt_cache_key": false
            }
          }
        ]
      }
    }
  }
}

Do not put the key in settings.json. Enter it in the provider's settings in the agent panel, where Zed stores it in the system keychain, or let Zed read it from an environment variable named after the provider, which here is XFIELD_API_KEY.

Hermes Agent

Run hermes model, choose the custom endpoint option, and enter https://xfield.cloud.brutal.ai/api/v1, your key and a model id. Or edit ~/.hermes/config.yaml:

~/.hermes/config.yaml
YAML
model:
  default: deepseek-v4.1-flash-1017-uncensored
  provider: custom
  base_url: https://xfield.cloud.brutal.ai/api/v1
  key_env: XFIELD_API_KEY

Anything else

If a tool takes an OpenAI base URL, it will almost certainly work. Run through this list:

  • Base URL is https://xfield.cloud.brutal.ai/api/v1. It already ends in /v1. Do not append /chat/completions unless the tool asks for a full endpoint.
  • Anthropic-only tools take https://xfield.cloud.brutal.ai/api instead; the SDK adds /v1/messages.
  • Key goes in as a bearer token. If the tool asks for an "OpenAI API key", that is the field.
  • Model id must match exactly, including -uncensored. List them with GET https://xfield.cloud.brutal.ai/api/v1/models.
  • Limits. Tell the tool the context and output size from models, or it will guess low.
  • Images only for qwen-3.8-27b-uncensored. Turn vision off elsewhere.
  • Test first. A JSON list from this call means the URL and key are right:
Shell
curl -sS "https://xfield.cloud.brutal.ai/api/v1/models" -H "Authorization: Bearer $XFIELD_API_KEY"

Tools with dedicated pages: Claude Code, Codex, OpenCode.