Game Tutorial
Teach the next meaningful action at the moment it becomes relevant, while keeping full rules available as reference.
1. Inspect the project and request
Identify the engine or runtime, entry flow, run command, input-binding system, tutorial state storage, and string organization. Determine whether the user requested an audit or implementation; an audit ends with findings and a proposed form.
2. Audit the first-time experience
Play from a clean state and learn only from what a new player can see, try, and receive as feedback. If the game cannot run, trace the displayed entry flow from source and mark the result inferred.
Inventory controls, goal and failure conditions, score meaning, and non-obvious rules. Classify every gap:
- First-action essential — needed before the player can take a meaningful action.
- Contextual — teach when the mechanic or input first becomes relevant.
- Reference — useful but not needed in the active flow.
- Discoverable — safe experimentation and feedback already teach it.
Teach the first three classes in their proper channel. Leave discoverable behavior alone so necessary guidance is not buried under the obvious.
3. Choose the form
Use the least intrusive form the existing game can support:
- Guided first steps. Attach one prompt to an existing level, wave, encounter, or event when the action first matters.
- Staged first run. Script the opening of an existing mode to expose mechanics in sequence, for example by delaying pressure until the first action succeeds. Preserve the normal mode and ask before changing level or encounter design.
- Contextual hints. Observe state, show a hint on first relevance, and dismiss it after the player performs the action.
- Instruction screen. When gameplay has no usable hook, show the first-action essentials from step 2 before play on a screen the player can dismiss.
Provide a reachable “How to Play” or equivalent reference for rules that do not belong in the active flow. Reference depth may match the game; it is not constrained by the first-action gate.
4. Build the state and guidance
Represent at least unseen, in progress, completed, and skipped. Persist completed and skipped states, and provide an intentional reset or replay path when the project has settings or help UI.
- Show the actual bound key, button, gesture, or control for the active input.
- Present one required action at a time and dismiss it on successful action rather than on a timer.
- Make skipping available in one action without removing access to reference help.
- Keep prompts correct when the active input changes among supported capabilities.
- Store player-facing tutorial text in the existing string catalog; if none exists, centralize it with stable identifiers instead of scattering literals through scenes.
- Preserve the normal pace after the guided or staged portion completes.
5. Verify
Test these paths:
- clean state → guidance appears in order and completes;
- skip → active guidance stops, the state persists, and reference help remains usable;
- completed state → the tutorial does not block a returning player;
- reset or replay → guidance can be intentionally reached again;
- each supported input → prompts show actions the capability can perform;
- smallest supported viewport → prompt and instruction text remains legible;
- interruption or restart during progress → state resumes or restarts deliberately.
If the project cannot run, deliver these as verification steps and mark implementation behavior unverified.
Constraints
- Teach the core loop in the gate and defer depth progressively. A game can have many valid rules without being a design defect.
- Keep tutorial prompts skippable and reference help reachable.
- Propose changes to level, encounter, difficulty, or core rules and apply them only with user approval.