emberlm

Official Node.js SDK for EmberLM. Run versioned prompts from your production code, evaluate LLM responses, and ship AI with confidence.

Install

npm install emberlm

Requires Node.js 18+ (for native fetch).

Quick start

Get an API key from Settings → API Keys in the dashboard. Keys start with pk_live_.

import { Client } from "emberlm";

const client = new Client({ apiKey: "pk_live_..." });

const result = await client.run("summarize_docs", {
  variables: { document: docText },
});

console.log(result.text);
console.log("passed_evals:", result.passed_evals);
console.log("confidence:", result.confidence);
console.log("cost_usd:", result.cost_usd);
console.log("latency_ms:", result.latency_ms);

API

new Client(options)

new Client({
  apiKey: "pk_live_...",
  baseUrl: "https://emberlm.dev", // optional
  timeoutMs: 60_000,              // optional, defaults to 60s
  fetch: globalThis.fetch,        // optional, pass your own
});

client.listPrompts()

Returns every prompt in the API key's workspace.

const { prompts } = await client.listPrompts();
// prompts: Array<{ id, name, description, model, current_version, tags }>

client.getPrompt(name)

Fetch a single prompt by name. Returns the prod-tagged version if one exists, otherwise the latest.

const prompt = await client.getPrompt("summarize_docs");
// { id, name, version, tag, system_prompt, user_prompt, model, ... }

client.run(promptName, options?)

Run a saved prompt. Variables are substituted into {{placeholders}}. All workspace eval rules are applied to the output. The run is persisted and counts toward the workspace's analytics.

const result = await client.run("classify_ticket", {
  variables: { body: ticket.body },
  model: "claude-haiku-4-5", // optional override
});

// result:
// {
//   run_id, prompt, version, model,
//   text,
//   input_tokens, output_tokens, total_tokens,
//   cost_usd, latency_ms,
//   passed_evals, confidence, evals: [...],
//   error,
// }

client.eval(options)

Evaluate an arbitrary response against the workspace's active eval rules. Useful when the response was generated by a different SDK or model.

const { passed, confidence, results } = await client.eval({
  response: llmResponse,
  prompt: userPrompt,               // optional
  variables: { name: "Jane" },      // optional
  ruleIds: ["rule-id-1", "rule-id-2"], // optional, defaults to all active
});

Errors

Any non-2xx response throws an EmberLMError:

import { Client, EmberLMError } from "emberlm";

try {
  await client.run("missing_prompt");
} catch (err) {
  if (err instanceof EmberLMError) {
    console.error(err.status, err.message);
  }
}

| Status | Meaning | |---|---| | 401 | Missing or invalid API key | | 403 | API key's plan does not permit the SDK | | 404 | Prompt not found in the workspace | | 429 | Monthly call limit reached, or rate limited |

Rate limits

100 requests / minute per API key.

License

MIT