Stream message

curl -N -X POST https://api.vouchmark.com/v1/chat/chat_aBcD.../stream \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"message": "Summarise my monitoring alerts from last week"}'

POST

chat

{chatId}

stream

curl -N -X POST https://api.vouchmark.com/v1/chat/chat_aBcD.../stream \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"message": "Summarise my monitoring alerts from last week"}'

Same behaviour as Send message, but the AI reply is streamed back token-by-token for real-time rendering. The user message and the final AI message are still persisted to the chat.

Requires a positive wallet balance. Billed at ₦100 per 1,000 tokens consumed by the request, and rate limited to 10 AI requests per minute.

Path parameters

string

required

The chat session ID.

Body

string

required

Your message, 1–4000 characters.

curl -N -X POST https://api.vouchmark.com/v1/chat/chat_aBcD.../stream \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{"message": "Summarise my monitoring alerts from last week"}'

Response format

The response is sent with Content-Type: application/x-ndjson. The body is a sequence of newline-delimited JSON objects, one per line. Each object has the shape { "type": <string>, "data": <any> }. This is NDJSON, not Server-Sent Events, so read it with a streaming body reader rather than EventSource:

{"type":"text","data":"Based on"}
{"type":"text","data":" your monitoring"}
{"type":"text","data":" data from the past week..."}
{"type":"grounding_sources","data":{"groundingChunks":[{"web":{"uri":"https://...","title":"FIRS"}}]}}
{"type":"finish","data":{"complete":true}}

Event types

`type`	`data`	Description
`text`	string	A text fragment of the reply. Concatenate all `text` events for the full message.
`tool_call`	`{ toolName, args }`	The AI invoked a tool (e.g. web search) mid-response. Informational.
`grounding_sources`	`{ groundingChunks, searchEntryPoint? }`	Sources used to ground the answer. Emitted once, before `finish`, only when present.
`finish`	`{ complete: true }`	The stream is complete.
`error`	`{ message }`	An error occurred mid-stream. The connection then closes.

Client example

const response = await fetch(`/v1/chat/${chatId}/stream`, {
  method: "POST",
  headers: {
    Authorization: `Bearer ${token}`,
    "Content-Type": "application/json",
  },
  body: JSON.stringify({ message: "..." }),
});

const reader = response.body!.getReader();
const decoder = new TextDecoder();
let buffer = "";

while (true) {
  const { done, value } = await reader.read();
  if (done) break;

  buffer += decoder.decode(value, { stream: true });
  const lines = buffer.split("\n");
  buffer = lines.pop() ?? "";

  for (const line of lines) {
    if (!line.trim()) continue;
    const event = JSON.parse(line);
    if (event.type === "text") {
      appendToUI(event.data);
    }
  }
}

Overview

Auth

Companies

Verifications

Onboarding (KYB)

Widget Embed (Runtime)

Trust Badge

Monitoring

Investigations

Disputes

Vouch AI

Compliance

Wallet & Billing

Developers

Path parameters

Body

Response format

Event types

Client example

​Path parameters

​Body

​Response format

​Event types

​Client example

Path parameters

Body

Response format

Event types

Client example