First Chat

Your first conversation with an LLM using Perspt.

Prerequisites

  • Perspt installed (see Quick Start)

  • An API key for any provider

Step 1: Set Your API Key

Choose your provider and set the environment variable:

export OPENAI_API_KEY="sk-..."

export ANTHROPIC_API_KEY="sk-ant-..."

export GEMINI_API_KEY="..."

# No key needed, just ensure Ollama is running
ollama serve

Step 2: Launch Perspt

perspt

Or with a specific model:

perspt chat --model gpt-5.2

Step 3: The TUI Interface

You’ll see the Perspt TUI:

┌─────────────────────────────────────────────────────────────┐
│  Perspt v0.5.0 - gpt-5.2                     Tokens: 0     │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│                                                             │
│                   Welcome to Perspt!                        │
│            Your Terminal's Window to the AI World           │
│                                                             │
│                                                             │
├─────────────────────────────────────────────────────────────┤
│  > Type your message here...                                │
└─────────────────────────────────────────────────────────────┘

Step 4: Send a Message

Type your message and press Enter:

> What is the capital of France?

The response will stream in real-time with markdown rendering.

Step 5: Continue the Conversation

Keep chatting! The conversation history is maintained:

> And what's the population?

The population of Paris is approximately 2.1 million in the city
proper, and about 12 million in the metropolitan area.

Step 6: Save Your Conversation

Use the /save command:

> /save my_chat.md

Or with automatic timestamp:

> /save

Step 7: Exit

Press Esc or Ctrl+C to exit cleanly.

Key Bindings Reference

Key

Action

Enter

Send message

Esc

Exit application

↑/↓

Scroll chat history

Page Up/Down

Fast scroll

Ctrl+C

Force exit

Tips

  1. Markdown works: Use code, bold, and lists in your prompts

  2. Long responses: Scroll up to see earlier content

  3. Token tracking: Watch the token counter in the header

  4. Model switching: Use perspt chat --model <name> for different models

Next Steps