Function handleRun

packages/ai-sandbox-cloudflare/src/runner.ts:164–188

Stream the agent's chunks to the response as NDJSON, one object per line.

(
  req: IncomingMessage,
  res: ServerResponse,
  resolveAdapter: (input: ResolveAdapterInput) => AnyTextAdapter,
)

/** Stream the agent's chunks to the response as NDJSON, one object per line. */
async function handleRun(
  req: IncomingMessage,
  res: ServerResponse,
  resolveAdapter: (input: ResolveAdapterInput) => AnyTextAdapter,
): Promise<void> {
  const parsed: unknown = JSON.parse(await readBody(req))
  const request = parseContainerRunRequest(parsed)
  res.writeHead(200, {
    'content-type': 'application/x-ndjson',
    'cache-control': 'no-cache',
  })
  // The DO appends each line to its durable run-log; here we are the producer,
  // so we surface a mid-stream failure as a terminal RUN_ERROR line the DO will
  // append + finish on, never a silently truncated stream.
  try {
    for await (const chunk of runAgent(request, resolveAdapter)) {
      res.write(`${JSON.stringify(chunk)}\n`)
    }
  } catch (error) {
    const message = error instanceof Error ? error.message : String(error)
    res.write(`${JSON.stringify({ type: EventType.RUN_ERROR, message })}\n`)
  } finally {
    res.end()
  }
}

/**
 * Start the in-container harness runner: a `node:http` server with `GET /health`