# Streaming logs (/sdk/logs)



Job logs stream over Server-Sent Events (SSE), not a request/response endpoint,
so they are intentionally **not** an SDK function. Streaming is two steps: mint
a short-lived log token with the SDK, then open the SSE URL directly.

## 1. Mint a log token [#1-mint-a-log-token]

<SdkSignature name="getBuildLogToken" />

```ts
import { getBuildLogToken } from "@harmont/cloud";

const { data: tok, error } = await getBuildLogToken({
  client,
  path: { org: "acme", pipeline: "ci", number: 42 },
});
if (error) throw new Error(error.error.code);
```

The token is short-lived and scoped to that build's logs.

## 2. Open the SSE stream [#2-open-the-sse-stream]

Open the log endpoint with the token as a query parameter and iterate the
response body. Note the log stream lives at `/v0/jobs/{jobId}/logs` — outside
the `/api/v0` base the other endpoints use:

```ts
const res = await fetch(
  `https://api.harmont.dev/v0/jobs/${jobId}/logs?token=${tok.token}`,
);
const reader = res.body!.getReader();
const decoder = new TextDecoder();
for (;;) {
  const { value, done } = await reader.read();
  if (done) break;
  process.stdout.write(decoder.decode(value));
}
```

List a build's jobs with `listJobs` to get each `jobId`. See the
[end-to-end example](/sdk/example) for the full create → poll → stream flow.

<Callout type="info">
  The log token is passed as a query parameter because `EventSource` and many
  SSE clients can't set custom headers. It is redacted from server-side request
  logs.
</Callout>