CSV handling

netlify/functions/agent-proxy.js

@@ -38,6 +38,49 @@ specific dollar amounts.`
 
 const MODEL = process.env.ANTHROPIC_MODEL || 'claude-sonnet-4-5'
 
+// File uploaded to the Anthropic Console / Files API that the agent should
+// consult on every call.
+const KNOWLEDGE_FILE_ID =
+  process.env.ANTHROPIC_KNOWLEDGE_FILE_ID || 'file_011Cawv1SPNaknuMmyDtM1S7'
+
+// ---------------------------------------------------------------------------
+// Module-scoped singletons (reused across warm Lambda invocations)
+// ---------------------------------------------------------------------------
+// Netlify keeps the function process alive between invocations as long as it
+// stays warm, so anything declared at module scope persists. We lazily build
+// the Anthropic client and download the knowledge file exactly once per
+// cold start, then reuse them for the lifetime of the container.
+
+let _client
+function getClient() {
+  if (_client) return _client
+  if (!process.env.ANTHROPIC_API_KEY) {
+    throw new Error('Server is missing ANTHROPIC_API_KEY')
+  }
+  _client = new Anthropic({
+    apiKey: process.env.ANTHROPIC_API_KEY,
+    // Required while the Files API is in beta.
+    defaultHeaders: { 'anthropic-beta': 'files-api-2025-04-14' },
+  })
+  return _client
+}
+
+// Cache the in-flight promise so concurrent requests during a cold start
+// share a single download instead of fetching the CSV N times.
+let _knowledgeFilePromise
+function loadKnowledgeFile() {
+  if (_knowledgeFilePromise) return _knowledgeFilePromise
+  _knowledgeFilePromise = (async () => {
+    const resp = await getClient().beta.files.download(KNOWLEDGE_FILE_ID)
+    return await resp.text()
+  })().catch((err) => {
+    // Reset on failure so the next request can retry.
+    _knowledgeFilePromise = undefined
+    throw err
+  })
+  return _knowledgeFilePromise
+}
+
 // ---------------------------------------------------------------------------
 // Handler
 // ---------------------------------------------------------------------------
@@ -68,12 +111,20 @@ export default async (req /*, context */) => {
     return new Response('Invalid station website URL', { status: 400 })
   }
 
-  if (!process.env.ANTHROPIC_API_KEY) {
-    return new Response('Server is missing ANTHROPIC_API_KEY', { status: 500 })
+  let client
+  let knowledgeFileText
+  try {
+    client = getClient()
+    // First call after a cold start awaits the download; subsequent calls
+    // resolve immediately from the cached promise.
+    knowledgeFileText = await loadKnowledgeFile()
+  } catch (err) {
+    return new Response(
+      `Failed to initialize agent: ${err.message || err}`,
+      { status: 500 },
+    )
   }
 
-  const client = new Anthropic({ apiKey: process.env.ANTHROPIC_API_KEY })
-
   const userMessage = `Here is a new public media station intake:
 
 - Submitter Name: ${userName}
@@ -93,7 +144,27 @@ Please draft the funding outlook brief described in your instructions.`
     model: MODEL,
     max_tokens: 10000,
     system: SYSTEM_PROMPT,
-    messages: [{ role: 'user', content: userMessage }],
+    messages: [
+      {
+        role: 'user',
+        content: [
+          {
+            type: 'document',
+            source: {
+              type: 'text',
+              media_type: 'text/plain',
+              data: knowledgeFileText,
+            },
+            title: 'Funder knowledge base (CSV)',
+            citations: { enabled: true },
+            // Cache the tokenized CSV on Anthropic's side for ~5 minutes
+            // so repeat requests pay ~10% of input-token cost for it.
+            cache_control: { type: 'ephemeral' },
+          },
+          { type: 'text', text: userMessage },
+        ],
+      },
+    ],
   }
   if (AGENT_ID && AGENT_ID !== 'YOUR_AGENT_ID_HERE') {
     streamArgs.agent_id = AGENT_ID