AI Help Chat

AI Help Chat is the in-app assistant for your local mock API server. Use it for troubleshooting, explanations, and guided slash-command changes with a preview/apply safety step.

Before you chat

1. For commands that create servers, complete activation first via Activate and license.
2. Open Settings → AI and choose provider and model.
3. For cloud models, save an API key for the selected model.
4. Offline provider does not require an API key.
5. In chat, ask a concrete question (example: “Why did this request return Not Mocked?”).

Best first prompts

• “Why did this request return Not Mocked?”
• “Help me create a GET /health endpoint with JSON response.”
• “Set failure simulation for /users to 20% with a 503 response.”
• “Explain why my passthrough route is failing.”

Slash commands

Type / in the input to open command suggestions. Use mutating commands for guided changes and /explain when you only want guidance with no workspace edits.

• /new resets chat state and starts a new conversation.
• /create-server creates a new mock server.
• /create-endpoint creates an endpoint and lets you finish early with done or /preview.
• /set-failure sets delay/failure simulation for an endpoint.
• /set-mode sets endpoint mode (MOCK/PASSTHROUGH/DISABLED).
• /explain ... asks for non-mutating guidance.

If activation is not complete, /create-server will open the activation prompt instead of applying changes.

Preview and apply workflow (safe path)

1. Start a slash command (for example /create-endpoint).
2. Answer clarifying questions in chat.
3. Use done or /preview when you are ready.
4. Review the preview plan summary and changed values carefully.
5. Click Apply or Cancel (or type apply/cancel).

No partial apply is performed when preview validation fails. If workspace state changes after preview, regenerate preview before applying.

When to use chat vs manual UI

• Use chat when you want guided setup, troubleshooting, or a preview before editing.
• Use manual UI when you already know the exact field you want to change quickly.
• Use /explain first when you want to understand behavior before editing.

Natural language endpoint creation

If you type a request like “create new endpoint repo details”, the assistant can infer a likely method/path, ask for confirmation, and then build a preview you can apply.