Getting started

Quick start

From nothing to your first completed testing round in about fifteen minutes. This walkthrough uses Claude Code; the same steps work in any supported client — see Setup & connection for the others.

What you'll do

1. Get a key
2. Connect a client
3. Confirm your workspace connection
4. Set up your product
5. Recruit a tester
6. Run one round
7. Read the result

Step by step

1. Get a license key. Sign up at www.squat.pro — the free key takes a minute and needs no card. Your key (it starts with sqk_) connects you to your workspace: the live methodology, your hosted squad member, and your ledger. One key works on up to three devices.

2. Install the SQUAT plugin in Claude Code.

   /plugin marketplace add ElucidateDigital/SQUAT-Plugin
   /plugin install squat@squat

3. Connect your key. Add two environment variables to your shell profile, then restart Claude Code:

   export SQUAT_SERVER_URL="https://api.squat.pro/mcp"
   export SQUAT_LICENSE_KEY="sqk_..."

   Paste the server address whole — scheme and /mcp included. If your organization runs its own SQUAT server, substitute the address your operator gives you.

4. Verify the connection. Ask Claude: “Check my SQUAT connection.” If it reports the workspace identity and member list, you're connected.

5. Confirm hosted protection. Connected sessions protect personas, scenarios, moderator questions, dialogue, results, reports, and project-artifact files automatically. You do not create a workspace content key or recovery file before recruiting or running a panel.

6. Orient with /start. The front door: say “/start” (or just “get started with SQUAT”) anytime to see your workspace status, your roster, and the most likely next moves as a menu. If anything needs fixing — such as no project or squad set up — the first option offered is the fix.

   If you want a practice run first, ask to “install the free SQUAT tutorial.” SQUAT previews the tutorial pack and asks before adding its practice persona, squad, scenario, and discussion guide to your workspace.

7. Set up your product. Say “set up a project for my app.” A project is the record of the product under test — what it is, who it's for, how to reach it (its target), and the core features that matter most. Setting it up first keeps your history attributable and gives every tester the context they need. This — not recruiting — is the first move.

8. Add a scenario. A scenario is the tester's task, stated as a goal in the user's own words — not click-by-click instructions. You can describe one yourself, or say “show me the starter scenarios” to import a ready-made one (like “first-visit orientation” or “core task completion”) from SQUAT's starter library.

9. Recruit or assemble your squad. Now choose who tests it. Say “recruit a tester” and SQUAT interviews you — one question at a time, usually 8–12 questions — about one real person who uses (or should use) your product: their goals, how they behave when something's confusing, their tech comfort, their voice. You can also draw on personas already in your roster, or mix both — each member joins this project's squad.

   Tip

   Already have a persona document — research notes, a marketing avatar, a support-ticket profile? Say so. SQUAT will offer to import it directly instead of running the full interview. See Recruiting testers.

10. Run the round. Say “run a UAT round.” Your AI plays your squad member attempting the scenario in character — driving your real app through a connected browser tool if one is available — then a separate evaluator scores the transcript on what actually happened and submits the result to the ledger.

11. Read the result. Say “how did the round go?” You'll get a plain-language readout: what got tested, what passed, what broke, and what to look at first — plus the ledger table your history will build on from here.

What just happened

You now have a workspace with one hosted squad member, one project, and one completed round in the ledger. Everything from here compounds:

• Each future round adds a point to the trend — pass/fail counts, regressions, findings.
• Your squad member accumulates learnings from each round and starts behaving like a returning user.
• Every finding your testers report becomes a tracked issue that SQUAT follows until it's verified fixed — and raises an alarm if it comes back.

Where to go next