Prompt#

The prompt keyword is one of AIScript's core AI integration features, allowing you to easily interact with large language models (LLMs) directly from your code.

Basic Usage#

The simplest form of the prompt keyword takes a string argument and returns the model's response:

let answer = prompt "What is the capital of France?";
print(answer); // The capital of France is Paris.

In this basic form, AIScript will use the default model configuration (typically OpenAI's model) to generate the response.

Advanced Configuration#

For more control over the prompt behavior, you can provide additional configuration options:

let detailed_answer = prompt {
    input: "Explain quantum computing",
    model: "claude-3.7",
    max_tokens: 500,
    temperature: 0.7,
    system_prompt: "You are a quantum physics expert"
};

Configuration Options#

Dynamic Prompts#

You can construct prompts dynamically using variables and string interpolation:

let topic = "quantum computing";
let detail_level = "beginner";
let response = prompt f"Explain {topic} at a {detail_level} level";

API Keys and Configuration#

To use the prompt feature, you need to configure your API keys. AIScript supports multiple LLM providers.

Environment Variables#

Set your API keys using environment variables:

# For OpenAI
export OPENAI_API_KEY="your_openai_key"

# For Anthropic Claude
export CLAUDE_API_KEY="your_claude_key"

Configuration File#

Alternatively, you can specify API keys in your project configuration file:

[ai.openai]
api_key = "YOUR_OPENAI_API_KEY"

[ai.anthropic]
api_key = "YOUR_CLAUDE_API_KEY"

Error Handling Not supported yet#

Like other AIScript operations, you can use error handling with the prompt keyword:

let response = prompt "Generate a complex response" |err| {
    print(f"Error occurred: {err}");
    return "Fallback response";
};

Best Practices#

• Be specific: The more specific your prompt, the better the response quality.
• Consider context limits: Each model has a maximum context window. Very large prompts may be truncated.
• Handle rate limits: API providers have rate limits. Consider implementing retry logic for production applications.
• Manage costs: Using LLMs incurs costs based on token usage. Monitor usage to avoid unexpected charges.

Examples#

Simple Question-Answering#

let question = "What is the capital of Japan?";
let answer = prompt question;
print(answer);

Generating Content with Specific Parameters#

let poem = prompt {
    input: "Write a haiku about programming",
    temperature: 0.9,
    max_tokens: 100
};
print(poem);

Using Different Models#

// Using OpenAI
let openai_response = prompt {
    input: f"Summarize this article: {article_text}",
    model: "gpt-4"
};

// Using Claude
let claude_response = prompt {
    input: f"Summarize this article: {article_text}" ,
    model: "claude-3.7"
};

Chaining Prompts#

let topic = prompt "Suggest a topic for a blog post";
let outline = prompt f"Create an outline for a blog post about: {topic}";
let introduction = prompt f"Write an introduction for a blog post with this outline: {outline}";

print("Topic:", topic);
print("Outline", outline);
print("Introduction:", introduction);

Notes and Limitations#

• The response from a prompt is always a string. If you need structured data, consider asking for specific formats like JSON in your prompt or parse the result.
• Network connectivity is required for the prompt keyword to function, as it makes API calls to LLM providers.
• API costs apply based on your provider's pricing model, typically based on the number of tokens processed.