Helicone

Helicone is request-side observability: it observes the LLM calls your application makes. PromptVault does not push prompt source to Helicone.

What It Does

The Helicone integration helps you keep Helicone credentials and prompt version metadata aligned with your application. Your app fetches a prompt through the SDK, then attaches prompt metadata to the outbound LLM request.

Configuration

For a custom Helicone API host, enter your own URL. Otherwise leave the default.

Test

Use Test after connecting to confirm the key is valid.

Suggested SDK Usage

const prompt = await pv.get("support-triage");

await openai.chat.completions.create(
  { model: prompt.model, messages: prompt.messages },
  {
    headers: {
      "Helicone-Property-Prompt-Slug": prompt.slug,
      "Helicone-Property-Prompt-Version": prompt.version,
    },
  },
);

This tags each Helicone request with the slug and version label served by PromptVault.