Skip to content
AskVault Docs
Try Free →

Official SDKs

Written by , Founder, AskVault · Reviewed by Aashiq
Last updated: · 3 min read

Current state

SDKs are in alpha:

  • JavaScript / TypeScript. Internal alpha; request access via support.
  • Python. Internal alpha; request access.
  • Other languages. Use HTTP client directly.

Public release planned Q3 2026 once API stability confirmed.

Using fetch (JavaScript) today

Without SDK:

const response = await fetch('https://api.askvault.co/v1/query', {
  method: 'POST',
  headers: {
    'Authorization': 'Bearer ak_live_xxx',
    'Content-Type': 'application/json',
  },
  body: JSON.stringify({
    workspace_id: 'ws_xxx',
    query: 'What is your refund policy?',
  }),
});
const data = await response.json();

Using requests (Python) today

import requests


response = requests.post(
    'https://api.askvault.co/v1/query',
    headers={'Authorization': 'Bearer ak_live_xxx'},
    json={'workspace_id': 'ws_xxx', 'query': 'What is your refund policy?'},
)
print(response.json())

What the SDKs will provide

When released:

  • Auth helper plus token refresh.
  • Retry logic with exponential backoff for 5xx and 429.
  • Streaming wrapper for /v1/query/stream.
  • HMAC computation for identity verification.
  • Webhook signature verification.
  • TypeScript types for full request/response shapes.

Sample SDK usage (alpha preview)

JavaScript:

import { AskVault } from '@askvault/sdk';
const client = new AskVault({ apiKey: 'ak_live_xxx' });


const response = await client.query({
  workspaceId: 'ws_xxx',
  query: 'What is your refund policy?',
});
console.log(response.answer);

Python:

from askvault import AskVault
client = AskVault(api_key='ak_live_xxx')


response = client.query(workspace_id='ws_xxx', query='What is your refund policy?')
print(response.answer)

About 90% less boilerplate than raw HTTP.

Alpha access

To request alpha:

  1. Email developer@askvault.co.
  2. Describe your use case.
  3. Receive access within 5 business days.

Roadmap

Beyond JS and Python:

  • Go. Q4 2026.
  • Ruby. Q4 2026.
  • PHP. TBD.
  • Java / Kotlin. TBD.

Community SDKs welcome; we'll list them once stable.

Limits

  • API rate same as raw HTTP.
  • SDK overhead. Under 5 ms per call.
  • Token storage. In-memory by default.
  • Retry attempts on 5xx. 3 with exponential backoff over 30 seconds.
  • Streaming chunks. Up to 100 per response.

Common pitfalls

Awaiting SDK release. Use HTTP client; no functional difference.

Old SDK version. Update regularly during alpha phase; API may evolve.

Token storage. Don't bundle into client-side code; SDK is server-side.

FAQ

Why not use OpenAI-compatible client?

API shapes differ. A direct SDK is more idiomatic.

Will the SDK be open-source?

Yes. MIT license planned.

Can I contribute?

Once public, yes via GitHub.

Was this page helpful?