Developer API

The Developer API is OpenAI-compatible. Point any OpenAI SDK at https://api.corebasehq.com/api/v1, pass a cb_live_ token, and it just works — chat completions stream in the standard chat.completion.chunk SSE format. The model has access to your tenant’s data sources, Unified Context, and access policies.

Base URL

https://api.corebasehq.com/api/v1

For self-hosted deployments, replace with your CoreBase domain.

What you can do

Chat completions

OpenAI-compatible streaming chat. POST /api/v1/chat/completions.

Sources

List the data sources your agents and chat can reach. GET /api/v1/sources.

Widget usage

Per-end-user usage and live quota counters for your embedded widget — build partner dashboards on it. GET /api/v1/usage.

Plan

Check the subscription tier attached to your token’s organization.

Quick example

Using the official OpenAI SDK:

from openai import OpenAI

client = OpenAI(api_key="cb_live_xxx", base_url="https://api.corebasehq.com/api/v1")

stream = client.chat.completions.create(
    model="corebase",
    messages=[{"role": "user", "content": "How many active customers last month?"}],
    stream=True,
)
for chunk in stream:
    print(chunk.choices[0].delta.content or "", end="", flush=True)

Or with curl:

curl -N -X POST https://api.corebasehq.com/api/v1/chat/completions \
  -H "Authorization: Bearer cb_live_xxx" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "corebase",
    "messages": [{"role": "user", "content": "How many active customers last month?"}],
    "stream": true
  }'

LLM model

You bring your own LLM provider API key — Anthropic, OpenAI, Gemini, Groq, Mistral, Azure OpenAI, OpenRouter, or Ollama. See LLM keys for the panel setup. The model field in your request is accepted but ignored; the actual model is whichever provider key you’ve marked as default.

Versioning

The Developer API is versioned at the URL prefix:

https://api.corebasehq.com/api/v1/chat/completions

We avoid breaking changes within a major version. When v2 ships, v1 stays online for at least 12 months with deprecation warnings in the Sunset header.

Rate limits

Each cb_live_ token has its own per-minute, per-day, and per-month limits configured at issuance. Hit a limit and you get 429 Too Many Requests with Retry-After.

Errors

HTTP	Meaning
400	Body schema invalid (e.g. `stream: false`)
401	Missing or invalid token
403	Token lacks the required permission
409	No LLM provider key configured for the organization
429	Rate limit exceeded
5xx	Transient server-side issue — safe to retry with exponential backoff

The body is JSON for any non-2xx response. 4xxs carry a stable detail.error string you can branch on; 5xxs carry a request ID for support.

Authentication

Create tokens, scope permissions, rotate secrets.

Chat endpoint

Full request / response spec for POST /api/v1/chat/completions.

Endpoints

Webhooks

Developer API

Base URL

What you can do

Chat completions

Sources

Widget usage

Plan

Quick example

LLM model

Versioning

Rate limits

Errors

Next

Authentication

Chat endpoint

​Base URL

​What you can do

Chat completions

Sources

Widget usage

Plan

​Quick example

​LLM model

​Versioning

​Rate limits

​Errors

​Next

Authentication

Chat endpoint

Base URL

What you can do

Quick example

LLM model

Versioning

Rate limits

Errors

Next