Basic Usage¶

Day-to-day operations with Perspt.

Starting Perspt¶

# Default: starts chat TUI
perspt

# With specific model
perspt chat --model gpt-5.2

# From anywhere
export PATH="$PATH:/path/to/perspt/target/release"
perspt

Chat Interface¶

┌─────────────────────────────────────────────────────────────┐
│  Perspt v0.5.0 - gpt-5.2                     Tokens: 1,234 │
├─────────────────────────────────────────────────────────────┤
│                                                             │
│  User: What is recursion?                                   │
│                                                             │
│  Assistant: Recursion is a programming technique where a    │
│  function calls itself to solve a problem by breaking it    │
│  down into smaller subproblems.                            │
│                                                             │
│  ```python                                                  │
│  def factorial(n):                                          │
│      if n <= 1:                                             │
│          return 1                                           │
│      return n * factorial(n - 1)                            │
│  ```                                                        │
│                                                             │
├─────────────────────────────────────────────────────────────┤
│  > _                                                        │
└─────────────────────────────────────────────────────────────┘

Sending Messages¶

Type your message in the input area
Press Enter to send
Watch the streamed response with markdown rendering

Key Actions¶

Enter	Send message
Ctrl+J or Ctrl+Enter	Insert newline in input
PageUp / PageDown	Scroll message history (10 lines)
Ctrl+Up / Ctrl+Down	Scroll message history (1 line)
Mouse Scroll	Scroll message history (3 lines)
Ctrl+C or Ctrl+Q	Exit Perspt

Scrolling Behavior¶

The chat interface features intelligent auto-scroll:

During streaming: Automatically scrolls to show new content
Manual scroll up: Disables auto-scroll so you can read previous messages
Scroll to bottom: Re-enables auto-scroll when you reach the end

The TUI uses virtual scrolling to handle very long conversations efficiently, rendering only the visible portion of the message history.

Saving Conversations¶

> /save
Saved to: conversation_2024-12-23_15-30-00.md

> /save my_chat.md
Saved to: my_chat.md

Switching Models¶

# In a new session
perspt chat --model claude-opus-4.5

# In chat (if supported)
> /model gemini-3-flash

Working with Code¶

Perspt renders code blocks with syntax highlighting:

> Write a Python function to reverse a string

Here's a concise solution:

```python
def reverse_string(s: str) -> str:
    return s[::-1]

# Usage
print(reverse_string("hello"))  # "olleh"
```

Multi-turn Conversations¶

The chat maintains context:

> Write a Calculator class

```python
class Calculator:
    def add(self, a, b): return a + b
    def subtract(self, a, b): return a - b
```

> Add multiply and divide methods

```python
class Calculator:
    def add(self, a, b): return a + b
    def subtract(self, a, b): return a - b
    def multiply(self, a, b): return a * b
    def divide(self, a, b):
        if b == 0: raise ValueError("Cannot divide by zero")
        return a / b
```

Token Usage¶

The header shows cumulative token usage:

Input tokens: Your prompts
Output tokens: AI responses
Total: Running sum for cost estimation

Exit¶

Press Esc or Ctrl+C for a clean exit.