Skip to main content

​
Overview

The getPrompt function retrieves prompts stored in the Lucidic platform, enabling centralized prompt management across your applications.

​
Syntax

await lai.getPrompt({
  promptName: string,
  variables?: Record<string, any>,
  cacheTtl?: number,  // seconds; -1 = forever; 0 = no cache
  label?: string,     // default 'production'
}): Promise<string>

​
Parameters

​
promptName
string
required
The name/identifier of the prompt to retrieve.
​
variables
Record<string, any>
Optional variables to substitute (e.g., { name: 'Alice' } replaces {{name}}).
​
cacheTtl
number
default:"300"
Cache TTL in seconds. Use -1 to cache forever, 0 to disable cache.
​
label
string
default:"'production'"
Prompt label/version to fetch.

​
Returns

Returns a Promise that resolves to the prompt content as a string.

​
Examples

​
Basic Usage

// Retrieve a prompt
const prompt = await lai.getPrompt({ promptName: "customer_support_greeting" });

// Use with OpenAI
const response = await openai.chat.completions.create({
  model: "gpt-4",
  messages: [
    { role: "system", content: prompt },
    { role: "user", content: userMessage }
  ]
});

​
With Cache Control

// Force fresh fetch (bypass cache)
const latestPrompt = await lai.getPrompt({ promptName: "dynamic_prompt", cacheTtl: 0 });

// Cache for 5 minutes (default 300)
const cachedPrompt = await lai.getPrompt({ promptName: "static_prompt", cacheTtl: 300 });

​
Error Handling

try {
  const prompt = await lai.getPrompt("missing_prompt");
} catch (error) {
  console.error("Prompt not found:", error);
  // Fall back to default prompt
  const prompt = "You are a helpful assistant.";
}

​
Prompt Versioning Pattern

// Use environment-specific prompts
const environment = process.env.NODE_ENV || "development";
const promptName = `assistant_prompt_${environment}`;

const prompt = await lai.getPrompt(promptName);

​
Multiple Prompts

// Load multiple prompts
const [systemPrompt, userPrompt, errorPrompt] = await Promise.all([
  lai.getPrompt("system_instructions"),
  lai.getPrompt("user_template"),
  lai.getPrompt("error_handler")
]);

// Use in conversation
const messages = [
  { role: "system", content: systemPrompt },
  { role: "user", content: userPrompt.replace("{{input}}", userInput) }
];

​
Common Patterns

​
1. Template Replacement

const prompt = await lai.getPrompt({
  promptName: "email_template",
  variables: {
    customerName: "John Doe",
    product: "Premium Plan",
    date: new Date().toLocaleDateString()
  }
});

​
2. Prompt Caching Strategy

class PromptManager {
  private cache = new Map<string, { prompt: string; timestamp: number }>();
  private cacheTTL = 5 * 60 * 1000; // 5 minutes

  async getPrompt(name: string): Promise<string> {
    const cached = this.cache.get(name);
    
    if (cached && Date.now() - cached.timestamp < this.cacheTTL) {
      return cached.prompt;
    }
    
    // Force fresh fetch for expired cache
    const prompt = await lai.getPrompt({ promptName: name, cacheTtl: 0 });
    this.cache.set(name, { prompt, timestamp: Date.now() });
    
    return prompt;
  }
}

​
3. Fallback Prompts

async function getPromptWithFallback(
  primaryName: string,
  fallbackName: string
): Promise<string> {
  try {
    return await lai.getPrompt({ promptName: primaryName });
  } catch (error) {
    console.warn(`Primary prompt ${primaryName} not found, using fallback`);
    try {
      return await lai.getPrompt({ promptName: fallbackName });
    } catch (fallbackError) {
      // Ultimate fallback
      return "You are a helpful AI assistant.";
    }
  }
}

​
Integration Example

class AIAssistant {
  private openai: any;

  async initialize() {
    const OpenAI = (await import('openai')).default;
    await lai.init({ instrumentModules: { OpenAI } });
    this.openai = new OpenAI();
  }

  async respond(userInput: string, context?: string): Promise<string> {
    // Get prompts from platform
    const systemPrompt = await lai.getPrompt("assistant_system");
    const contextPrompt = context 
      ? await lai.getPrompt("assistant_with_context")
      : "";

    const messages = [
      { role: "system", content: systemPrompt },
    ];

    if (context && contextPrompt) {
      messages.push({
        role: "system",
        content: contextPrompt.replace("{{context}}", context)
      });
    }

    messages.push({ role: "user", content: userInput });

    const response = await this.openai.chat.completions.create({
      model: "gpt-4",
      messages
    });

    return response.choices[0].message.content;
  }
}

​
Best Practices

​
1. Naming Conventions

// Use descriptive, hierarchical names
await lai.getPrompt("support/greeting/enterprise");
await lai.getPrompt("analysis/financial/quarterly");
await lai.getPrompt("generation/email/followup");

​
2. Error Recovery

// Always have a fallback strategy
async function safeGetPrompt(name: string, defaultPrompt: string): Promise<string> {
  try {
    return await lai.getPrompt(name);
  } catch (error) {
    await lai.createEvent({
      description: "Prompt fetch failed",
      result: `Failed to get ${name}: ${error.message}`
    });
    return defaultPrompt;
  }
}

​
3. Performance Optimization

// Pre-load critical prompts
async function preloadPrompts(promptNames: string[]) {
  const prompts = await Promise.all(
    promptNames.map(name => lai.getPrompt({ promptName: name }))
  );
  
  return Object.fromEntries(
    promptNames.map((name, i) => [name, prompts[i]])
  );
}

// Usage at startup
const criticalPrompts = await preloadPrompts([
  "system_instructions",
  "error_handler",
  "greeting_message"
]);

​
Notes

  • Prompts must be created in the Lucidic dashboard first
  • Caching improves performance for frequently used prompts
  • Prompt names are case-sensitive
  • Network errors will throw exceptions

​
See Also