Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/jorgeurtubiam-ship-it/Gulin_ia/llms.txt

Use this file to discover all available pages before exploring further.

AI presets let you define named bundles of ai:* settings that can be selected from the model dropdown in any AI block. Instead of re-entering an API key, base URL, or model name every time you switch providers, you save each configuration as a preset and switch between them with a single click.

​
What Is a Preset?

A preset is a named entry in presets.json whose key starts with ai@. Each preset carries any combination of ai:* configuration fields along with optional display metadata. When you select a preset from the model dropdown, its fields are merged over the global ai:* defaults defined in settings.json. The special field "ai:*": true in a preset acts as a reset flag — it clears all inherited ai:* values before applying the preset’s own fields, ensuring a clean slate instead of an unintended merge.
Name your presets with the ai@ prefix followed by a short, descriptive slug — for example ai@ollama-local, ai@gpt4o, or ai@deepseek. The text after ai@ is what users see in the dropdown when display:name is not set.

​
Built-In Presets

GuLiN ships two presets in defaults/presets/ai.json:

ai@global

The Global default preset (order -1). It uses "ai:*": true to inherit whatever ai:* values are configured in your settings.json. Selecting this preset effectively applies your top-level AI settings with no overrides.

ai@gulin

The Gulin Proxy – gpt-5-mini preset (order 0). Uses the GuLiN-hosted proxy with gpt-5-mini, 4 000 max tokens, and a 60-second timeout. No API key is required for this preset.
The raw definitions from defaults/presets/ai.json: 
{
    "ai@global": {
        "display:name": "Global default",
        "display:order": -1,
        "ai:*": true
    },
    "ai@gulin": {
        "display:name": "Gulin Proxy - gpt-5-mini",
        "display:order": 0,
        "ai:*": true,
        "ai:apitype": "",
        "ai:baseurl": "",
        "ai:apitoken": "",
        "ai:name": "",
        "ai:orgid": "",
        "ai:model": "gpt-5-mini",
        "ai:maxtokens": 4000,
        "ai:timeoutms": 60000
    }
}

​
Preset Fields

Every preset is a JSON object whose keys are drawn from the following set:
FieldTypeDescription
display:namestringHuman-readable label shown in the model dropdown.
display:ordernumberSort position in the dropdown (lower numbers appear first).
ai:*booleanWhen true, clears all inherited ai:* values before applying this preset.
ai:apitypestringAPI protocol: openai-chat, openai-responses, or google-gemini.
ai:baseurlstringBase URL for the provider endpoint.
ai:apitokenstringAuthentication token for the provider.
ai:modelstringModel identifier passed in each request.
ai:namestringOverride the display name inside the AI block header.
ai:orgidstringOrganization ID (OpenAI and compatible).
ai:apiversionstringAPI version string (Azure and versioned endpoints).
ai:maxtokensnumberMaximum tokens per response.
ai:timeoutmsnumberRequest timeout in milliseconds.
ai:proxyurlstringHTTP/HTTPS proxy for AI requests.
ai:fontsizenumberProportional font size for the AI chat view.
ai:fixedfontsizenumberFixed (monospace) font size in the AI chat view.

​
Creating a Custom Preset

Custom presets live in presets.json inside your GuLiN config directory. Open it with: 
wsh editconfig presets.json
1

Open presets.json

Run wsh editconfig presets.json in any GuLiN terminal. If the file does not exist yet, GuLiN creates an empty one for you.
2

Add your preset object

Add a top-level key using the ai@ prefix. Include "ai:*": true to start from a clean slate, then set only the fields you need.
3

Save and select

Save the file. Your preset appears immediately in the model dropdown inside any AI block — no restart required.

​
Example: Local Ollama Preset

{
    "ai@ollama-local": {
        "display:name": "Ollama – Llama 3.2 (local)",
        "display:order": 1,
        "ai:*": true,
        "ai:apitype": "openai-chat",
        "ai:baseurl": "http://localhost:11434/v1",
        "ai:apitoken": "ollama",
        "ai:model": "llama3.2",
        "ai:maxtokens": 4000,
        "ai:timeoutms": 120000
    }
}

​
Example: OpenAI GPT-4o Preset

{
    "ai@gpt4o": {
        "display:name": "OpenAI – GPT-4o",
        "display:order": 2,
        "ai:*": true,
        "ai:apitype": "openai-chat",
        "ai:baseurl": "https://api.openai.com/v1",
        "ai:apitoken": "$ENV:OPENAI_API_KEY",
        "ai:model": "gpt-4o",
        "ai:maxtokens": 8000,
        "ai:timeoutms": 60000
    }
}
GuLiN supports $ENV:VARIABLE_NAME syntax in config values. Use it to read API keys from environment variables instead of storing them in plain text. You can also supply a fallback with $ENV:VAR:fallback-value.

​
Setting the Default Preset

The gulinai:defaultmode setting in settings.json controls which AI mode is active when a new AI block opens. It defaults to gulinai@balanced. To change the default to your custom preset: 
wsh setconfig gulinai:defaultmode="ai@ollama-local"
Alternatively, set ai:preset to point to your preferred preset as the global default: 
wsh setconfig ai:preset="ai@gpt4o"

​
Feature Flag: GulinApp Builder

Setting feature:gulinappbuilder to true in settings.json enables the experimental GulinApp builder UI, which provides a visual interface for scaffolding AI-powered terminal applications. 
wsh setconfig feature:gulinappbuilder=true
The GulinApp builder is an experimental feature. Its API surface and behavior may change between releases.

Build docs developers (and LLMs) love

Get started for freeTalk to us