Skip to main content

Documentation Index

Fetch the complete documentation index at: https://mintlify.com/jkh2/Primordial-Sim/llms.txt

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

The AI Lab Partner supports four providers through a unified adapter. You can switch providers at any time from within the AI Lab panel — your settings persist in localStorage between sessions.

​
Provider comparison

ProviderEndpointDefault modelAPI key required
OpenAIapi.openai.com/v1/chat/completionsgpt-4oYes
Anthropicapi.anthropic.com/v1/messagesclaude-sonnet-4-20250514Yes
xAIapi.x.ai/v1/chat/completionsgrok-3-miniYes
Custom / LocalAny OpenAI-compatible URLhttp://localhost:1234/v1/chat/completionsNo

​
Setting up a provider

1

Open the AI Lab panel

Press L or click the star icon (⚛) in the top-right corner of the screen.
2

Expand provider settings

Click the AI Provider Settings header at the top of the panel. The settings section collapses and expands on click.
3

Select your provider

Use the Provider dropdown to choose OpenAI, Anthropic, xAI, or Custom / Local. The model field updates automatically to the default for that provider. If you select Custom / Local, an additional Endpoint URL field appears.
4

Enter your API key

Paste your API key into the API Key field. The field uses password masking. For local models, leave this field empty — no key is required.
5

Set the model (optional)

The Model field is pre-filled with the provider’s default. Change it to any model your account has access to (e.g. gpt-4-turbo, claude-opus-4-5, grok-3).
6

Save and test

Click Save Settings to persist your configuration, then click Test Connection to verify the provider responds. The status line below the buttons shows the result.

​
Custom / Local models

The Custom / Local option accepts any URL exposing an OpenAI-compatible chat completions endpoint. Compatible servers include:
  • LM Studio — default URL: http://localhost:1234/v1/chat/completions
  • Ollama — default URL: http://localhost:11434/v1/chat/completions
  • vLLM — configure to match your server’s port
  • text-generation-webui — enable the OpenAI extension
Set the Endpoint URL to your server’s address, leave the API Key field blank, and set the Model field to the model name your server expects.
Local models require no API key and work fully offline (aside from the simulation itself, which has no network requirements). This is a good option if you want to experiment without cloud API costs or for environments without internet access.

​
Format differences handled transparently

The callProviderAPI adapter normalizes provider-specific request and response formats so the rest of the system does not need to care which provider is active. Anthropic uses the Messages API:
  • System prompt is passed as a separate top-level system parameter
  • Requires the anthropic-version and anthropic-dangerous-direct-browser-access headers
  • Returns a content array of content blocks; the adapter joins block text to produce a single string
OpenAI, xAI, and Custom/Local use the chat completions format:
  • System prompt is prepended as a {role: "system"} message in the messages array
  • The Authorization: Bearer <key> header is used for authentication
  • Returns choices[0].message.content as the response text

​
Settings persistence

Settings are saved to localStorage under the key primordial-ai-config as a JSON object containing provider, apiKey, model, and customUrl. They are loaded automatically when the page opens. Your API key never leaves your browser except in direct API calls to your chosen provider.
Some cloud AI providers block direct browser-to-API requests with CORS headers. Anthropic explicitly supports direct browser access with the anthropic-dangerous-direct-browser-access header, which Primordial includes automatically. OpenAI and xAI generally allow direct browser requests. If you encounter CORS errors with a custom endpoint, you may need to configure your local server to include Access-Control-Allow-Origin: * response headers.

Build docs developers (and LLMs) love

Get started for freeTalk to us