LatentKit JavaScript SDK

Official JavaScript/TypeScript client for the canonical LatentKit /v1 API.

• Requires Node 18+ or another runtime with standard fetch, ReadableStream, and AbortController.
• ESM-only package.
• Route-based by design: do not pass model, provider, route, or policy in SDK requests. The API key's assigned published route selects the provider/model at runtime.

Install

npm install @latentkit/sdk

Quickstart

import { LatentKit } from '@latentkit/sdk';

// apiKey/baseUrl fall back to LATENTKIT_API_KEY / LATENTKIT_BASE_URL when omitted.
const client = new LatentKit();
const response = await client.chat.create({
  messages: [{ role: 'user', content: 'Say hello from LatentKit.' }],
  max_tokens: 100,
  response_profile: 'balanced',
});

console.log(response.content);

Configuration

| Option | Env fallback | Notes | | --- | --- | --- | | apiKey | LATENTKIT_API_KEY | Explicit value wins; throws if neither is set | | baseUrl | LATENTKIT_BASE_URL | Defaults to https://ai.latentkit.com | | toolSlug / toolVersion | — | Sent as X-LK-Tool-Slug / X-LK-Tool-Version for attribution | | appId | — | X-LK-App-ID for account-scoped IDE tokens |

Control plane

With an account IDE access token (lkia_, from device login), client.control.* exposes read/inspect helpers:

await client.control.listApps();
await client.control.listLogs({ tenantId: 't_123', days: 7 });
await client.control.usageSummary('t_123', 30);
await client.control.listRoutes('t_123');
await client.control.simulateRoute('pol_1', 't_123', { endpoint: 'chat' });

Errors are LatentKitApiError with .status, .code, and .request_id.

Route-based requests

The SDK is not model-based. Your application sends the task body and optional response_profile; LatentKit resolves the provider/model from the API key's assigned route. model in a response is reporting metadata for the route that won, not a request field.

SDK calls reject route-control keys such as model, provider, route, and policy, including inside extra_body.

Inspect the connected route

client.me.retrieve() returns a typed MeResponse with the connected app and workspace, credits, assigned route, ordered route models, and latest winning request when those fields are available:

const context = await client.me.retrieve();

console.log(context.policy?.name, context.policy?.model_count);
for (const model of context.policy?.models ?? []) {
  console.log(model.rank, model.provider, model.model);
}
console.log(context.latest_request?.model);

The latest winner is request activity, not a permanently selected model. Runtime SDK credentials cannot change the assigned route.

Errors

import { LatentKit, LatentKitApiError } from '@latentkit/sdk';

const client = new LatentKit({ apiKey: process.env.LATENTKIT_KEY! });

try {
  await client.chat.create({
    messages: [{ role: 'user', content: 'hello' }],
  });
} catch (error) {
  if (error instanceof LatentKitApiError) {
    console.error(error.status, error.code, error.body);
  }
}

Streaming

import { LatentKit } from '@latentkit/sdk';

const client = new LatentKit({ apiKey: process.env.LATENTKIT_KEY! });

for await (const event of client.chat.stream({
  messages: [{ role: 'user', content: 'Count from one to five.' }],
})) {
  if (event.event === 'error') throw new Error(JSON.stringify(event.data));
  if (event.isDone) break;
  console.log(event.data);
}

Response profiles

Pass response_profile directly to ask the assigned policy for a speed/depth tradeoff:

import { LatentKit } from '@latentkit/sdk';

const client = new LatentKit({ apiKey: process.env.LATENTKIT_KEY! });

const response = await client.chat.create({
  messages: [{ role: 'user', content: 'Think through this carefully.' }],
  response_profile: 'deep',
});

Allowed values are fast, balanced, and deep. The assigned policy controls whether request overrides are allowed and which routes are eligible for each profile.

Chat, image, and embeddings

import { LatentKit } from '@latentkit/sdk';

const client = new LatentKit({ apiKey: process.env.LATENTKIT_KEY! });

const chat = await client.chat.create({
  messages: [{ role: 'user', content: 'Write one sentence about LatentKit.' }],
});

const image = await client.image.generate({
  prompt: 'A clean product icon on a white background',
  size: '1024x1024',
});

const vectors = await client.embeddings.create({
  input: ['hello world'],
  dimensions: 256,
});

Agent sessions

import { LatentKit } from '@latentkit/sdk';

const client = new LatentKit({ apiKey: process.env.LATENTKIT_KEY! });

const session = await client.agents.sessions.create({
  task: 'Inspect the repo and explain the auth flow',
  workspace_root: '/workspace',
  permission_mode: 'workspace-write',
});

const queued = await client.agents.sessions.run(String(session.id));
console.log(queued);

See docs/latentkit-coder-api.md for the full agent session shape.

Modalities

import { LatentKit } from '@latentkit/sdk';

const client = new LatentKit({ apiKey: process.env.LATENTKIT_KEY! });

await client.embeddings.create({ input: ['hello world'], dimensions: 256 });
await client.image.generate({ prompt: 'A clean product icon', size: '1024x1024' });
await client.speech.create({ input: 'Hello from LatentKit.', voice: 'alloy' });
await client.transcription.create({ audio: { base64: '...' }, language: 'en' });
await client.translation.create({ audio: { base64: '...' }, target_language: 'en' });
await client.video.generate({ prompt: 'A short product scene', duration_seconds: 4 });

Queue requests support every endpoint accepted by POST /v1/queue and an optional idempotency key:

await client.queue.create({
  endpoint: 'embeddings',
  payload: { input: ['hello world'] },
  idempotencyKey: 'import-row-42',
});

Queue support is enqueue-only. There is no public per-job result endpoint, so workflows that need results should use their host job runner and call a synchronous SDK resource from that job.

Supported resources

• client.chat.create(...)
• client.chat.stream(...)
• client.completions.create(...)
• client.completions.stream(...)
• client.vision.create(...)
• client.vision.stream(...)
• client.embeddings.create(...)
• client.image.generate(...)
• client.speech.create(...)
• client.transcription.create(...)
• client.translation.create(...)
• client.video.generate(...)
• client.queue.create(...)
• client.agents.sessions.create(...)