This app runs entirely standalone — no AI vendor, no external account, no login. Everything works the same whether opened inside claude.ai, saved as a plain file and opened in any browser, or hosted internally on a corporate intranet.

One app, four industry streams. Maritime & Shipping (treated as one, since they share publications, regulators, and client base), Offshore & Energy, Navy & Defence, and Cruise & Passenger — each with its own publications and its own search vocabulary. Each has a Worldwide pack, plus a panel per region you choose, plus optional panels for individual vessel and asset segments. Clients are ticked against one or more industries, and can additionally be watched by IMO number and stock ticker.

Every client needs at least one industry. It is required, and it scopes the vocabulary their search uses — the main lever you have over result relevance. It does not change how many requests a run makes (a client search is one query either way), so treat it as a precision setting rather than a speed one. Clients imported without an industry are flagged on the Clients tab with a control to set a whole group at once. Region defaults to Worldwide and is optional: these industries are international, and region does not affect a client's news search.

Clients, Regions, and Opportunities save to this browser's own local storage — nothing is sent anywhere for that.

Industry News, Region News, and Client Watch fetch public RSS feeds and Google News directly from the browser, through a CORS proxy (browsers can't read most feeds directly without one). That proxy layer is the app's main reliability risk — the three public fallback services aren't SLA-backed, so they can rate-limit or go down without warning, especially across a 30-request refresh of every client. Set PRIVATE_PROXY_BASE near the top of the file to your own Cloudflare Worker URL (deploy kambuddy-proxy-worker.js, included alongside this app — free tier, a few minutes to set up) to get a relay that's yours alone and isn't competing with everyone else's traffic. The public proxies stay in as an automatic fallback either way.

Priority scoring (the 0–100 score, "why it matters," recommended actions, and the services suggested on each story) is computed instantly in the browser using a transparent point-based rule set — see below. No AI call is involved, so it never fails, times out, or depends on any external service.

Data is stored per-browser, per-device — it does not sync between your laptop and phone, or between browsers, automatically. Use the backup export below to move it, or the file sync feature (Chrome/Edge) to keep a linked copy elsewhere.