feat(worker): add weekly job scrape worker with portal integrations, scoring, and Twenty CRM sync

apps/worker/wrangler.jsonc

@@ -0,0 +1,40 @@
+{
+  "$schema": "./node_modules/wrangler/config-schema.json",
+  "name": "jobyssey-scrape",
+  "main": "src/index.ts",
+  "compatibility_date": "2026-09-21",
+  "compatibility_flags": ["nodejs_compat"],
+  "observability": { "enabled": true, "head_sampling_rate": 1 },
+  // Scoring is one Claude call per posting; CPU is mostly JSON + HTML parsing.
+  "limits": { "cpu_ms": 300000 },
+  // Profile + rubric + queries are bundled as text so the Worker scores against
+  // the same files /rank reads locally. scripts/bundle-profile.ts copies them
+  // into src/profile/ before every deploy; `wrangler deploy` is the sync.
+  "rules": [{ "type": "Text", "globs": ["**/*.md"], "fallthrough": true }],
+  "workflows": [
+    {
+      "name": "weekly-scrape",
+      "binding": "WEEKLY_SCRAPE",
+      "class_name": "WeeklyScrape",
+      // Cloudflare cron: weekdays are 1=SUN..7=SAT, so name the day. 13:00 UTC = 09:00 EDT / 08:00 EST.
+      "schedules": ["0 13 * * MON"]
+    }
+  ],
+  "d1_databases": [
+    {
+      "binding": "POSTINGS",
+      "database_name": "jobyssey-postings",
+      "database_id": "c1df7f34-ed2b-4c74-b6af-10ef15c2db5d"
+    }
+  ],
+  "browser": { "binding": "BROWSER" },
+  "vars": {
+    "TWENTY_BASE_URL": "https://crm.prad.nu",
+    "TWENTY_ASSIGNEE_ID": "6fa482f3-cebd-4bfa-a096-d454697afc54",
+    "AI_GATEWAY_URL": "https://gateway.ai.cloudflare.com/v1/54c3f12c7eb138d1c426bcdbddb9d6fe/jobyssey/anthropic",
+    "SCORING_MODEL": "claude-sonnet-4-5",
+    // Hard cap on Claude calls per run; the rest of the pool is reported, not scored.
+    "MAX_SCORED_PER_RUN": "100"
+  }
+  // Secrets (wrangler secret put): TWENTY_API_KEY, ANTHROPIC_API_KEY, RUN_TOKEN
+}