Get help and choose your next project

Outcome and prerequisites

You can ask for useful technical help without losing the lesson context or sharing secrets. You can also define an intermediate milestone small enough to finish. Use this lesson whenever you get stuck, even if you have not completed the whole course.

Package the actual problem

A helper cannot reliably infer your Explorer hierarchy, script version or test mode from a screenshot of a character standing still. Gather:

• The current lesson ID and title
• The exact step you are attempting and what you expected
• What actually happened, with the complete first relevant Output error
• Whether this is a solo Test, server view, client view, or a multi-client session
• The script type and full Explorer path for every relevant script
• The smallest complete code needed to reproduce the issue
• The device/input and changes you made from the lesson

Use the help-context template. Copy it into a private note, or use the course site's context-copy control if available. Read the resulting text before sending it anywhere. A generated context block is a convenience, not evidence that the helper can see your Studio session.

Ask ChatGPT with files when useful

Attach the current Markdown lesson plus the relevant .luau files and paste the exact error. If you changed the code in Studio, export or copy that current version; the original download does not show your modifications. A screenshot of Explorer helps with object paths, while text is better for code and errors.

Ask for a diagnosis and the smallest justified change before requesting a rewrite. For example:

I am on lesson 07. Coin1 awards once in solo Test, but neither player gets a point in Server & Clients. Here are Collectibles, the Explorer hierarchy and the first Output error. Explain which assumption fails, then propose the smallest fix and a two-client verification test.

Do not paste authentication cookies, passwords, recovery codes, Open Cloud API keys, private account details or another person's private messages. Replace unnecessary personal identifiers with placeholders. These lessons do not require third-party credentials.

Evaluate an AI-generated fix

Check whether the proposed API exists in the current official reference. Confirm the suggested Script or LocalScript belongs where the answer says. Read the entire change for unexpected network calls, external module loads or destructive actions. Keep a known-good file before replacing code.

A plausible answer can still misunderstand replication, a character's lifetime or the exact version of your example. Test one change at a time. If the code now works, reproduce the original failure case plus a normal case and a negative case. For a boost, test eligible near, ineligible near, and eligible far away; success alone does not show that authorization works.

Avoid accepting a whole new framework to fix one nil lookup. If a helper proposes a dependency, ask what problem it solves and what new setup it introduces. You can add external tooling after you understand the Studio workflow.

Intermediate milestone

You are at a useful intermediate point when you can explain and demonstrate:

1. A clear game loop with a start, feedback and an end
2. Server-owned scoring and eligibility, with clients responsible for presentation and input intent
3. Two clients observing consistent state, including death, rejoining or late joining where applicable
4. Usable keyboard, touch and gamepad interactions
5. A narrow saving design with explicit failure behavior and an isolated test environment
6. A repeatable bug report, a performance measurement and a safe release checklist

This does not mean you have built production-grade persistence, monetization, matchmaking or anti-cheat. Those deserve separate focused work and appropriate review.

Choose one next extension

For the obby, add a timer and a personal best. First define when the clock starts, what invalidates a run and how a server verifies completion. For co-op, add two randomized coin layouts without changing the round-state contract. For art, replace one greybox object with a vetted asset and measure whether readability or performance improves.

Write acceptance criteria before coding: “Two clients can finish three rounds in a row, a player joining during Results sees the correct phase, and no old coins survive into the next round.” That is more useful than “make it polished.”

Troubleshooting and exercise

If help stays vague, shrink the reproduction. Start a fresh Baseplate and bring over only the failing mechanic. If the issue disappears, add dependencies one at a time. If every answer disagrees, use the current API reference and a minimal runtime experiment to resolve the disputed behavior.

Fill in the help template now for a bug you deliberately create, such as renaming Coins to Tokens without updating the HUD. Ask someone else whether they can reproduce it from the report. Then restore the correct name.

Sources and navigation

Verified 2026-10-03: Roblox API reference, debugging, Core Curriculum, Luau reference. AI-help and privacy checklists are original workflow recommendations.

Previous: Family playtests · Return to: Start here · Related: Help context, Project map