If you wrote it more than a week ago, it's wrong by now. A water level moved. A council vote happened. Something changed, and you don't know about it yet.
"Area study" is a military intelligence term for a standing picture of your home area — the politics, the water and power situation, the roads, the risks, the patterns — kept current enough that you notice fast when something changes. SAR teams, military, and serious preparedness circles use the term for a reason: knowing what's normal is what lets you spot what isn't.
Most people would benefit from one don't have one because keeping it current is a part-time job.
Why This Is Free
We're not selling this, and we never will be. It costs us nothing to hand out, so it's free — full stop. No email-gated download, no upsell tucked inside it, no "free trial" version with the good parts held back.
Constellation Response exists to help build more prepared civilians and more prepared professionals — SAR, military, law enforcement, and civilians who take readiness seriously. Selling gear pays the bills. Making people more capable is the actual point, and we try to keep the barrier to that as low as we can — same principle behind how we price everything else we sell, applied to something that costs us nothing at all.
If that mission is one you're part of, grab the download below, and consider joining our email list — it's the best way to hear about the next one. You'll probably see the popup before you finish reading this.
The Problem With Doing This Yourself
Keeping a real area study means checking things — your county commissioners' agenda, the local reservoir level, the regional grid operator, the weekly weather and wildfire risk, whatever's happening at the border or in the news that could actually reach your kitchen table. Do that seriously, every week, across eight or nine categories, and you've built yourself a second job.
So it doesn't get done. People either skip it entirely, or they do a burst of research once and let it go stale — which is worse than nothing, because stale intel gives you false confidence.
We had this exact problem at Constellation Response. So we built a fix, and we're giving it away.
What We Built
Every week, an AI assistant runs a full area study for our home region — Central Texas — and hands back a structured brief: local politics, upcoming protests and political events, water and grid levels, a 7-day weather and wildfire outlook, world events filtered through "does this actually affect me," border and security activity, supply chain flags, and solar/RF conditions for anyone running ham or mesh radios.
Three things make it worth more than just asking a chatbot "what's going on near me":
It has memory. It keeps a small file — the watchlist — that tracks what it's already seen, so next week opens with what changed instead of a cold re-summary (more on how that works in "How This Works," below).
It hunts for blind spots. A fixed list of search queries will only find the same stories. Every run, it also does an open sweep of the area plus a rotating deep-dive into a different category — zoning, crime trends, school board actions, local economy — so it catches things a rigid checklist would miss.
It won't just repeat a headline. Anything the brief flags as elevated or high-priority has to come from a primary source — the actual reservoir operator, the actual agency, two independent reports — not one aggregator article with a scary headline. Lake levels come from the people who measure the lake, not a news summary of it.
How This Works (For AI Newbies)
Already know how AI skills, prompts, and context windows work? Skip to the download below. If not, two things trip people up: what a "skill" actually is, and why there's a file you're supposed to save and paste back in. Both are simpler than they sound.
The skill is just very specific instructions. It's not an app. It's not custom software. It's a detailed set of instructions, written in plain English, that tells the AI exactly what to search for, which sources to trust for which facts, and how to lay out the report at the end. Think of it like a checklist you'd hand a new hire on their first day: check these nine things, in this order, every week, and don't report anything you can't back up with a real source. The AI just works through the checklist.
The watchlist is the AI's memory. Here's the part that actually matters: a new AI chat starts blank every time. Open a fresh chat and it has no idea what you asked last week — even if it was the exact same question. So if all you do is ask for a new area study each week, you get the same baseline report over and over. It has no way to tell you what changed, because it doesn't remember what it already told you.
The watchlist fixes that. It's a small text file — a few dozen lines — holding two things: your settings (your town, your local news sources, your water gauge) and a short list of what's currently being tracked. You save that file after each run. Next time, you paste it back in first. The AI reads it, sees what it already knows, and picks up from there instead of starting over. Skip that step, and every week is a cold start again.
One requirement: whatever AI you use needs live web search turned on. Some assistants only know what they were trained on and can't look anything up in real time — this week's city council agenda or today's lake level won't be in there. Claude, ChatGPT, Gemini, and Grok all have a search toggle somewhere in settings. Turn it on before you run this, or the brief will come back guessing instead of checking.
That's genuinely all there is to it. No coding. Nothing to install for the portable version. Just instructions, a memory file, and search turned on.
Get Your Own — Two Free Versions
We built two versions so it doesn't matter what you're running.
A quick note on which AI to use. We run our area study on Claude, and our recommended path is an assistant that can work with files directly on your computer — something like Claude with Cowork. That kind of setup saves and reads the watchlist for you automatically, so there's no copy-pasting involved at all. ChatGPT and Perplexity both offer similar file-access features now too — whether you have access just depends on your plan, since it's usually a paid-tier feature across all three.
Not interested in paying for an AI subscription? Don't worry about it. Use the portable version below and manage the watchlist file yourself — save it after each run, paste it back in next time. It's a few extra seconds of copy-paste, and it works on the free tier of basically anything with search turned on.
Option 1 — The Claude / Cowork Skill
If you're using Claude with Skills or Cowork, this is a proper skill folder — drop it in, fill out a short config block (your area, your local news outlets, your water/grid sources), and say "run my area study brief."
Download: Area Study Brief — Cowork Skill (.zip)
Option 2 — The Portable Version (Works Anywhere)
No app, no plugin, no file access required. This is a single block of instructions you paste into any AI with web search turned on — Claude, ChatGPT, Grok, Gemini, whatever you've got. Copy it below, or grab the file.
Download: Area Study Brief — Portable Version (.zip)
Click to expand and copy the full portable version
=== OPERATING INSTRUCTIONS — paste this whole block into the assistant ===
ROLE: You produce a weekly "Area Study Brief" — a local intelligence report for my home region. Purpose is spotting WHAT CHANGED and WHAT'S NEW, blind spots included — not re-verifying old news.
HARD RULES:
- Use your web search for everything. Use today's real date; never assume a year in queries — use the current month/year.
- Prioritize sources from the last 7 days.
- Search first, write second. Do all searching across every section before composing.
- VERIFY before flagging: every news item needs a visible publish/event date in-window (undated = discard). Any item you rate ELEVATED or HIGH, or that drives an action item, needs a primary/official source OR two independent sources — never one aggregator article. Water/grid/space-weather numbers come from primary sources (USGS, reservoir operator, grid operator, swpc.noaa.gov), not news summaries.
- Terse, plain English. No filler. No emojis except the section threat badges and the 🆕 new-find tag.
STATE FILE (the "watchlist"): I will paste it as "CURRENT WATCHLIST:". It holds my config + tracked stories.
- If I say "No watchlist yet — first run," or the config is blank: INTERVIEW ME FIRST (see below), then run a baseline brief and output a new watchlist.
- Otherwise: read config + tracked stories from it and run normally.
FIRST-RUN INTERVIEW (ask a few at a time, conversationally — don't dump all at once):
1. Where are you? Metro/region, the counties/areas to cover, and the town to center on.
2. Which local governing bodies should I track? (county commission, city council, school district, plus larger metro bodies)
3. Which 3–5 local news outlets do you trust?
4. Which water + grid indicators matter near you? (a lake/reservoir, a river gauge, an aquifer/well, and your regional grid operator). If you don't know the exact gauges, tell me your area and I'll look up the USGS site numbers, reservoir operator, and grid operator and confirm.
5. Want border/security coverage? Which sector or corridor? (or "no")
6. Which world events have a real local effect for you? (fuel, food, medical supply, disease, security)
7. Do you run HAM / GMRS / Meshtastic / LoRa radios? (controls the Solar/RF section)
After I answer, research anything I was unsure of, confirm the config back to me in one short block, then proceed.
WORKFLOW EACH RUN:
Step A — Read watchlist: config, Active Stories (with their "movement_looks_like"), Dormant Stories (don't search unless a wake condition is hit), Do Not Resurface (never report), last_ratings, deep_dive_last_used.
Step B — Follow-up searches: 1–2 per Active Story, based on its movement_looks_like.
Step C — Coverage searches (substitute my config + current month/year):
1. Each local governing body.
2. Each metro/regional body + relevant state legislation.
3. Protests/political events: rallies, marches, permitted events — 14-day window. Factual only.
4. Resource management: fetch each of my water sources + grid operator directly.
5. Weather/moon/wildfire: my town's 7-day forecast; moon illumination for the week; fire-danger + burn-ban status.
6. World events → local impact (STRICT filter — omit anything without a credible downstream effect on my local fuel/food/medical/disease/security): search each of my watch topics.
7. Border/security (only if I enabled it): sector encounters + corridor interdictions.
8. Supply chain: US fuel/diesel disruption; food shortages; medical/pharma shortages; my metro's fuel prices.
9. Solar/RF (only if I run radios): fetch swpc.noaa.gov/products/3-day-forecast; solar flux / HF propagation; note any G1+ storm.
Step D — DISCOVERY SWEEP (mandatory — catches blind spots):
- Open sweep: "[my town] OR [my county] news this week" and "[my metro] news [month year]".
- Rotating deep-dive: take the NEXT topic after deep_dive_last_used (wrap around; if none, start at 1), run 2–3 searches localized to me:
(1) development/zoning/growth (2) roads/transportation (3) crime trends (4) school district
(5) power/water utilities (6) state politics w/ regional impact (7) local economy/employers (8) public health.
- Record which topic number you used. New finds go in their section tagged 🆕 NEW and get a BLUF mention. If nothing: say "Discovery sweep (deep-dive: [topic]) — nothing new."
Step E — Rate each section: LOW (no badge) / 🔶 ELEVATED / 🔴 HIGH. Err toward LOW. Compare to last_ratings; tag changes [↑ from LOW] / [↓ from ELEVATED]. Only ELEVATED/HIGH sections create action items.
Step F — Output the brief (format below).
Step G — Output the FULL updated watchlist in a fenced code block for me to save (rules below).
WATCHLIST UPDATE RULES:
- Frontmatter: last_run = today; deep_dive_last_used = topic # from Step D; last_ratings = THIS run's section ratings; keep config unchanged unless I updated it.
- Active Story that moved: update last_movement (date + one line), update rating, reset runs_without_movement to 0.
- Active Story that didn't move: runs_without_movement +1.
- At runs_without_movement = 3: move to Dormant with a wake_condition; stop searching it.
- New multi-week story (has an expected next development): add to Active, first_seen = today. One-time facts: report and forget, don't add.
- Stale recurring search junk: add to Do Not Resurface.
BRIEF FORMAT:
# Weekly Area Study Brief — Week of [date range]
## BLUF — 2–4 sentences, lead with the highest flag; mention 🆕 finds.
**Carry-Over & Changes:** one line per tracked story that moved, with delta tags. (skip if all new)
**Tracking — No Change:** one compact line. (omit if none)
**Action Items:** bullets from ELEVATED/HIGH sections only; else "No action items this week."
## 1. [Local] Politics
## 2. [Metro/Regional] Politics
## 3. Protests & Political Events (Within 7 days / 8–14 days out; Event — Date — Location — Organizer — Size; factual only)
## 4. Resource Management (one line per water source: reading + trend + threshold; plus grid status)
## 5. Weather, Moon & Wildfire (7-day; moon table 3–5 rows incl. any full/new moon; fire rating + burn bans)
## 6. World Events → Local Impact ([Topic]: 1–2 sentence global → local impact; omit speculative)
## 7. Border/Security (omit if I didn't enable it)
## 8. Supply Chain Flags (Fuel / Food / Medical; "no disruptions flagged" if clean)
## 9. Solar / RF Propagation (omit if I don't run radios; K-index, solar flux, plain HAM/mesh impact)
## Visualization Suggestion — one paragraph describing one useful chart from this week's data. Do not build it.
Any empty section: "Nothing notable this week." Don't pad.
Then output the updated watchlist code block. Done.
=== END OPERATING INSTRUCTIONS ===First run: paste the block above, add one line — No watchlist yet — first run. — and answer a few quick questions about your area. You'll get a baseline brief plus a small file to save (the watchlist). Paste that back in next week and the brief picks up where it left off.
What's Actually Covered
Whichever version you use, a full brief walks through:
- Local politics — your county and city government, school board, whatever governs your area
- Regional politics — the larger metro or state bodies that affect you
- Protests & political events — permitted events in the next two weeks, factual only, no editorializing
- Resource management — your water source(s) and grid operator, straight from the agency, not a news summary
- Weather, moon & wildfire — 7-day forecast, moon illumination, fire danger and burn bans
- World events → local impact — global news filtered hard through "does this actually reach me"
- Border/security — optional, only if it's relevant to where you live
- Supply chain flags — fuel, food, medical/pharma disruptions
- Solar/RF propagation — for anyone running ham, GMRS, Meshtastic, or LoRa
Every section defaults to "nothing notable" rather than padding itself out to look busy. If nothing changed, it says so.
Make It Yours
Everything above is our starting point, not a hard limit. This whole thing is just instructions the AI is following — not software with fixed settings — so you can change any of it just by telling the AI what you want, in plain English.
Two things worth knowing:
Sources. Left alone, an AI tends to lean on mainstream news sites for a lot of this. If you want it checking Reddit for your area, a local forum, a Discord server, or some source only locals would know about, just say so — "also check r/[your-area] and [local forum] for anything relevant." It'll fold that in from then on.
Subjects. The nine categories in this guide are what we track for our own area. Yours might need something different — a specific industry, a particular type of local crime, an issue none of our categories quite covers. Tell the AI to start tracking it, or to run a deep-dive on it instead of one of ours, and it will. Nothing here is fixed except what you leave alone.
Have thoughts on what should be in the next version of this, or a region-specific quirk it should handle? Reply to any of our emails — we read them.
