AI Assistant

Rostyman includes a built-in AI assistant that helps you work with APIs faster — generate requests, explain responses, write test scripts, debug errors, and more.

Opening the AI Panel

• Press Ctrl+Alt+P (or Cmd+Opt+P on Mac) to toggle the AI panel
• Or click the AI tab in the right panel (Ctrl+J)

Setting Up

1. Add Your API Key to the Vault

The AI assistant uses API keys stored securely in the Vault (AES-256-GCM encrypted).

1. Open Settings > Vault (or click Vault in the status bar)
2. Add a new secret with your API key (e.g., key name: anthropic-key, value: your API key)
3. Enable the secret with the checkmark

2. Configure the AI Provider

1. Open the AI panel and click the Settings (gear) icon
2. Select your provider: Claude (Anthropic), OpenAI, or Ollama (local)
3. Choose a model from the dropdown — pricing is shown per 1M tokens
4. Select your vault secret from the API key dropdown

Ollama is free and runs locally — no API key needed. Just install Ollama and start it with ollama serve.

Providers & Models

Provider	Models	Pricing (input/output per 1M tokens)
Anthropic	Opus 4, Sonnet 4, Haiku 3.5	$0.80–$75
OpenAI	GPT-4o, 4o-mini, 4.1, 4.1-mini, 4.1-nano, o3, o4-mini	$0.10–$10
Ollama	Any installed model (llama3, mistral, etc.)	Free (local)

Quick Actions

When starting a new conversation, 7 quick action buttons appear:

Action	What It Does
Generate	Create an API request from a natural language description
Explain	Explain the current response in plain English
Tests	Generate rm.test() scripts from the current response
Fix	Analyze error responses and suggest fixes
Mock Data	Generate realistic sample JSON request bodies
Document	Create markdown documentation for the current request
Validate Schema	Compare response structure against expected schema

Conversational Chat

The AI assistant supports full multi-turn conversations:

• Type a message and press Enter to send (Shift+Enter for new line)
• The AI has context about your current request — method, URL, headers, body, response
• Conversations are saved to the database and persist across sessions
• After 20+ messages, older messages are automatically summarized to save context

Action Buttons

Code blocks in AI responses have contextual action buttons:

• JSON blocks → "Apply as Body", "Create Request"
• JavaScript blocks → "Apply as Test Script", "Apply as Pre-request Script"
• cURL/bash blocks → "Import as Request"

Click any button to apply the code directly to your current request or create a new one.

cURL Import

Paste a cURL command into the chat input — Rostyman auto-detects it and creates a new request tab with the parsed method, URL, headers, and body.

Managing Conversations

• Click the List icon to see all past conversations
• Rename a conversation by clicking the pencil icon
• Delete a conversation by clicking the trash icon
• Click + to start a new conversation

Tips

• The AI works best when you have a request open — it uses the active tab's context
• For Ollama, use llama3 or larger models for best results
• API keys never leave the main process — they're resolved from the vault at call time and never stored in the renderer
• Use "Validate Schema" to catch breaking API changes before they hit production