Surf Report Builder
Build the prompts that build your reports.
Surf Report Builder creates custom AI prompts that generate surf reports — tuned to your break, powered by live buoy and weather station data. Free.
No coding. Copy, paste, answer three quick questions.
The Builder runs in your favorite AI, not on this site.
How it works
Run the Builder once. Run your custom prompts before every session, forever.
Grab it below and drop it into your favorite AI. Answer three quick questions, including where and when you surf, and the Builder takes it from there.
The Builder hands you back prompts tuned for you and your spot — a Session Report prompt, a Long Range prompt, and an optional Tracking prompt that lets the system learn from your sessions over time. Save them somewhere you can find them.
Before every session, paste a saved prompt back into your AI. It pulls live data from NOAA, NWS, and FAA sources and gives you a fresh report tuned to your break, at this moment.
What you'll get
When the Builder finishes, you'll have three custom prompts saved and ready to run. Paste any of them into your AI any time you want a fresh report.
For your next session.
Run it before you paddle out. Pulls live buoy and tide data, checks the wind forecast, applies your break's local model, and tells you what to expect: size, shape, who the session favors, and what could make the call wrong.
For upcoming swell events.
Run it whenever you're planning ahead. A 3-to-21-day outlook focused on swell events that might reach your break — what's developing, when it might arrive, how confident the call is, and which events are worth planning around.
Score your reports. Build your own track record.
After you surf, log a quick score: how close reality felt to what the report predicted. Over time those scores build a personal track record the system uses to flag patterns and adjust confidence. Skip this one if you just want the reports.
Get the Builder
Surf Report Builder is a single piece of text. Copy it below, paste it into your favorite AI, and answer three quick questions.
Copy from here
Loading the Builder…Tip: copy from this page, not the Design Notes — PDFs sometimes break long prompts.
Curious how it works? Read the Design Notes.
Data sources
Most AI surf prompts can't check live data. They rely on whatever the model already knows — usually months out of date — or on web searches that pull bad buoy feeds or stale cached pages.
The Builder takes a different approach. When you run a custom prompt, it pulls from a small server we maintain that fetches data live from official sources before your AI sees it. No commercial forecasts. No interpretation. Just the actual numbers.
FAQ
Yes, it's free. There's no catch. It's a side project, no accounts, no data collection, no upsell.
ChatGPT (and other AIs) can't reliably find live buoy data, don't know your specific break, and tend to fill gaps by guessing. The Builder researches your break once, builds prompts that pull live data from NOAA, NWS, and FAA sources every time you run them, and is honest when data is missing instead of making something up.
You'll need an AI with live web access. Stick with paid tiers — free tiers tend to either make up data, fail partway through, or serve up day-old readings, none of which produces a useful report.
The paid AIs we recommend, in order:
Claude Pro — currently the most reliable. Handles the live data fetches cleanly and produces the most consistent reports.
ChatGPT Plus (Thinking mode only) — works, with two caveats. First, use Thinking mode, not Instant — Instant skips steps and produces abbreviated prompts. Second, paste the prompt rather than uploading it as a file. We've seen ChatGPT occasionally serve old cached data to file uploads that doesn't reproduce when the same prompt is pasted directly.
Perplexity — not currently recommended. Tends to serve cached results too old to be useful for live conditions.
AI tools change fast, so this guidance will evolve. If you try a different one and it works (or doesn't), the feedback form below is the right place to share that.
No. If you can copy text, paste it into your AI, and answer three brief questions, you can use the Builder.
Anywhere you can find them later. ChatGPT and Claude both have saved prompts features built in. You could also paste them into a Notes app, a Google Doc, an email or draft, or anywhere you already keep useful text.
Most coastal breaks have a usable buoy somewhere nearby — sometimes farther offshore than you'd expect. The Builder's research step looks for the closest usable stations and builds a local model that accounts for the distance and direction.
Breaks in genuinely instrument-poor areas (some remote stretches of coast) will produce reports with lower confidence, which the system labels honestly rather than hides.
That's up to you. Temps, crowds, board, skill level, what you're trying to work on — none of that fits in a forecast. Your Session Report tells you what's likely to happen at your break. You make the call.
Surfline and similar commercial sites mix raw data with editorial interpretation — ratings, predicted faces, "go" calls, stylized summaries. The Builder is designed to ignore all of that and rely only on the underlying official sources (NOAA, NWS, FAA).
Sometimes it will. Surf forecasting is genuinely hard — sensors go offline, models miss things, the ocean does what it wants. The Builder doesn't pretend otherwise. Every report names its sources, flags confidence honestly, and includes a "what could make this wrong" line so you know which way to be skeptical.
The optional Tracking prompt is built for this too. After you surf, you log how close reality felt to the report. Over time that history helps the system tune to your specific break.
Every report includes a confidence rating — High, Medium, Low, or Speculative. Speculative is the floor. It means the system couldn't confirm the core picture for your session window — the buoys, tide, or wind data either weren't reachable or were too old to trust as live conditions.
Most of the time, you'll see Speculative when the AI returns old cached data instead of live readings. When that happens, we think you should know rather than have the system pretend the data was fresh.
If you see Speculative, the simplest fix is to run the prompt again in a fresh chat session — sometimes the AI's cache releases on a new session. AI tools are improving fast, and the cache and fetch issues that produce Speculative reports today should get less common as these systems mature.
After you surf, you log a quick score: how close reality felt to what the report predicted, on a scale of -2 to +2. Over time those scores build a personal track record. The system uses that history to learn your break and tune future reports. It's a longer-term feature; the value compounds with every session you log.
Note: We recommend using the Tracking prompt with an AI that allows you to post it once and keep in memory, like a project or space. Otherwise, you'll have to paste the prompt every time you record a score.
The Builder doesn't send anything to us. From time to time, after a report runs, the system sends a small anonymous health check — what kind of report you ran, which AI you used, and whether it worked — so we can fix things that break across users. No name, no email, no IP, no personal details.
The optional Tracking prompt stores your scores in a Google Sheet that you create and own. Nothing is sent to us. Nothing is shared with other users.
This project is built and maintained by a Southern California surfer with 20+ years in the water. Started as a personal tool for tomorrow-morning decisions — surf or paddle? — and grew into this.
Feedback
Used the Builder? Tell us what worked, what didn't, what your local break needs, or what we got wrong. Local surfers see things we can't.