AI Configuration

Configure AI providers and prompts for AI-assisted features in A²D.

Overview

AI Configuration is available under Settings > AI Configuration. It controls how A²D uses large language models for features such as AI-generated tools, prompts, and assistance.

LLM Setup

Configure your AI provider:

• Provider — Select supported providers (e.g., OpenAI)
• API Key — Your provider API key (stored securely)
• Model — Model to use for generation
• Temperature — Sampling randomness, 0–2
• Max tokens — Ceiling on the generated response length

These settings power AI-assisted generation when creating tools, prompts, or other assets.

Use Test connection to verify the provider, key, and model before saving.

Reasoning models (GPT-5 and o-series)

OpenAI’s reasoning models — gpt-5 and its variants, plus the o1, o3, and o4 families — use a different request contract than gpt-4 and earlier. A²D detects these models automatically and adjusts the request for you:

• Max tokens is sent as the model’s completion-token limit rather than the legacy max_tokens parameter.
• Temperature is omitted, because these models accept only their default value.

No extra configuration is required — select the model and A²D handles the rest.

Reasoning models count internal reasoning against the completion-token limit, so a low Max tokens value can be consumed before any visible text is produced. If a reasoning model returns an empty response, raise Max tokens and retry.

AI Prompts

Manage the prompts A²D uses for AI features:

• View and edit prompt templates
• Customize behavior for your organization
• Adapt to different use cases or domains

Use Cases

• MCP Servers — AI-generated tool schemas from descriptions
• Agent Cards — AI-assisted skill definition
• Design Rules — AI-powered compliance suggestions

Next Steps

• MCP Servers — Create servers with AI assist
• Settings & Organization — All settings tabs

AI Configuration tailors A²D’s AI assistance to your needs.