@prism-lang/repl

Interactive REPL (Read-Eval-Print Loop) for the Prism programming language. Provides a command-line interface for experimenting with Prism code interactively.

📚 Full Documentation | 🚀 Getting Started | 🖥️ CLI Usage

Installation

The REPL is typically installed as part of the CLI package:

npm install -g @prism-lang/cli

Usage

Start the REPL by running:

prism
# or
prism repl

Features

• Interactive Evaluation: Execute Prism code line by line
• Multi-line Input: Support for functions and complex expressions
• Session Management: Track variables and execution history
• LLM Integration: Automatic provider detection from environment
• Confidence Formatting: Visual representation of confidence values
• Built-in Commands: Special REPL commands for session control

REPL Commands

All REPL commands start with : (colon):

| Command | Description | |---------|-------------| | :help | Show available commands and examples | | :vars | Display all variables in current session | | :clear | Clear session (reset all variables) | | :history | Show command history | | :stats | Display session statistics | | :llm | Show LLM provider status | | :exit | Exit the REPL (also Ctrl+D) |

Examples

Basic Usage

prism> 2 + 3
5 (number)

prism> x = 42 ~> 0.9
42 (~90.0%) (confident)

prism> greeting = "Hello, " + "World!"
"Hello, World!" (string)

prism> [1, 2, 3].map(n => n * 2)
[2, 4, 6] (array)

Confidence Values

prism> prediction = "rain" ~> 0.8
"rain" (~80.0%) (confident)

prism> conf = <~ prediction
0.8 (number)

prism> certain = 100 ~> 0.99
100 (~99.0%) (confident)

Functions

prism> add = (a, b) => a + b
[Function: add] (function)

prism> add(5, 3)
8 (number)

prism> greet = (name) => "Hello, " + name + "!"
[Function: greet] (function)

prism> greet("Alice")
"Hello, Alice!" (string)

LLM Integration

prism> response = llm("What is the capital of France?")
"The capital of France is Paris." (~85.0%) (confident)

prism> analysis = llm("Is this code secure?", { temperature: 0.2 })
"Based on my analysis..." (~75.0%) (confident)

Uncertain Control Flow

prism> value = 50 ~> 0.6
50 (~60.0%) (confident)

prism> uncertain if (value) {
...>   high { "High confidence" }
...>   medium { "Medium confidence" }
...>   low { "Low confidence" }
...> }
"Medium confidence" (string)

Session Management

The REPL maintains state between commands:

prism> :vars
Session Variables:
  x: 42 (~90.0%)
  greeting: "Hello, World!"
  prediction: "rain" (~80.0%)
  add: [Function: add]

prism> :stats
Session Statistics:
  Commands executed: 8
  Successful: 7
  Failed: 1
  Session duration: 2m 15s
  Average execution time: 125ms

prism> :clear
Session cleared.

prism> :vars
Session Variables:
  (empty)

Multi-line Input

The REPL automatically detects when more input is needed:

prism> function factorial(n) {
...>   if (n <= 1) {
...>     return 1
...>   }
...>   return n * factorial(n - 1)
...> }
[Function: factorial] (function)

prism> factorial(5)
120 (number)

Error Handling

The REPL provides helpful error messages:

prism> x + y
❌ Error: Variable 'y' is not defined

prism> 1 / 0
Infinity (number)

prism> llm()
❌ Error: Function 'llm' expects at least 1 argument, got 0

Environment Configuration

The REPL automatically detects LLM providers from environment variables:

export CLAUDE_API_KEY=your-claude-key
export GEMINI_API_KEY=your-gemini-key

prism
🚀 Prism REPL v1.0.0
🤖 LLM providers: claude, gemini, mock (default: claude)
Type :help for commands or :exit to quit

prism>

Without API keys:

prism
🚀 Prism REPL v1.0.0
⚠️  Only mock LLM provider available. Set CLAUDE_API_KEY or GEMINI_API_KEY for real AI integration.
Type :help for commands or :exit to quit

prism>

API Reference

PrismREPL Class

The main REPL implementation class:

class PrismREPL {
  constructor()
  
  // Evaluate a line of input
  async evaluate(input: string): Promise<REPLResult>
  
  // Register an LLM provider
  registerLLMProvider(name: string, provider: LLMProvider): void
  
  // Set the default LLM provider
  setDefaultLLMProvider(name: string): void
  
  // Get welcome message
  getWelcomeMessage(): string
  
  // Get current variables
  getVariables(): Record<string, any>
  
  // Clear session
  clearSession(): void
  
  // Get session statistics
  getStatistics(): SessionStats
}

REPLResult Interface

interface REPLResult {
  success: boolean;
  value?: any;
  error?: string;
  type?: string;
  shouldExit?: boolean;
}

Integration

The REPL is designed to be embedded in other applications:

import { PrismREPL } from '@prism-lang/repl';
import { MockLLMProvider } from '@prism-lang/llm';

// Create REPL instance
const repl = new PrismREPL();

// Set up providers
const provider = new MockLLMProvider();
repl.registerLLMProvider('mock', provider);
repl.setDefaultLLMProvider('mock');

// Evaluate code
const result = await repl.evaluate('x = 42 ~> 0.9');
console.log(result.value); // 42 (~90.0%)

Related Packages

• @prism-lang/core - Core language implementation
• @prism-lang/llm - LLM provider integrations
• @prism-lang/confidence - Confidence extraction utilities
• @prism-lang/validator - Validation toolkit
• @prism-lang/cli - Command-line interface

License

MIT