feat(intake): draft a listing from photos and a note (#223)

The client is a parameter rather than a module import, so every test passes a stub. A test that reaches the real API is a defect in the test: this runs on a route a stranger with a link can trigger, and each call costs money.

getAnthropicClient returns null rather than throwing when there is no key. An unconfigured environment is a working one, and the worker treats null exactly as it treats a failed call — one path rather than two.

parsed_output is guarded, not asserted. The SDK returns null there when the answer did not satisfy the schema, which is what a model replying in prose looks like; failing cleanly leaves the submission queued for a retry, where asserting would crash the worker mid-loop. Absent usage figures are treated as zero for the same reason: undercounting a cost is survivable, throwing away a draft that actually succeeded is not.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

backend/src/intake/anthropicClient.ts

@@ -0,0 +1,30 @@
+import Anthropic from '@anthropic-ai/sdk';
+
+/**
+ * The client, or null when there is no key.
+ *
+ * Null rather than a throw, because an unconfigured environment is a working
+ * one: submissions still arrive and wait undrafted. The worker treats null
+ * exactly as it treats a failed call, which keeps one path rather than two.
+ *
+ * Constructed once and cached. The SDK holds a connection pool, and building
+ * one per submission would be wasteful on a route a stranger can trigger.
+ */
+let cached: Anthropic | null = null;
+let resolved = false;
+
+export function getAnthropicClient(): Anthropic | null {
+  if (resolved) return cached;
+
+  const key = process.env.ANTHROPIC_API_KEY;
+  cached = key !== undefined && key.trim() !== '' ? new Anthropic({ apiKey: key }) : null;
+  resolved = true;
+
+  return cached;
+}
+
+/** Exposed for tests, which need a fresh decision per case. */
+export function resetAnthropicClient(): void {
+  cached = null;
+  resolved = false;
+}