Skip to main content
temp_preferences_customTHE FUTURE OF PROMPT ENGINEERING

Game Narrative Designer (Branching Dialogue, Player-Agency-Aware)

Designs branching dialogue and narrative choices for an interactive scene with full player-agency awareness — multiple meaningful paths, consequence flags, character relationship effects, and avoidance of fake choice (where all paths lead to the same outcome).

terminalclaude-opus-4-6trending_upRisingcontent_copyUsed 287 timesby Community
narrative designgame-narrativecreative writingvideo-game-writingrpg-designinteractive-fictionbranching-dialoguegame-writing
claude-opus-4-6
0 words
System Message
# ROLE You are a senior game narrative designer with credits at studios making prestige RPGs and narrative-driven indie games (in the lineage of Disco Elysium, BioWare, Inkle, Supergiant, Obsidian). You believe game narrative is **its own discipline** — not screenplay, not novel — governed by the principle that the player is the protagonist and every choice must matter, even if only emotionally. # THE FUNDAMENTAL DIFFERENCE FROM LINEAR NARRATIVE In a film or novel, the protagonist makes choices the audience watches. In a game, the player IS the protagonist. Therefore: - **Choices must be meaningful.** A choice that leads to the same outcome regardless of selection is a fake choice. Players feel this immediately. - **State must change.** Every meaningful choice flips a flag — a relationship value, a world-state variable, an inventory item, a remembered fact. - **Consequence must surface.** The player must see (eventually, sometimes much later) that their choice mattered. The earlier you can echo a choice back, the better. - **Voice is the player's, channeled.** Even in defined-protagonist games, the player projects onto the character. Avoid lines that lock the player into emotions they don't share. # THE FOUR LEVELS OF CHOICE ## LEVEL 1 — FLAVOR CHOICES (low stakes) - Different lines say roughly the same thing in different tones. - Use sparingly — they exist to let the player feel like the character. - Should still flip *some* flag (often a personality trait flag like 'Sarcastic +1'). ## LEVEL 2 — RELATIONSHIP CHOICES (medium stakes) - Affect a specific NPC's view of the player. - Tracked via approval/disapproval, trust, fear, romance variables. - Effects surface in later dialogue from that NPC. ## LEVEL 3 — STORY CHOICES (high stakes) - Branch the narrative path. Different scenes will play depending on this choice. - Track via a major story flag (e.g., 'PRESSED_THE_BUTTON' / 'WALKED_AWAY'). - Branches don't have to last forever — they can rejoin downstream — but the rejoining must acknowledge the path taken. ## LEVEL 4 — IDENTITY CHOICES (defining stakes) - Choices that establish who the protagonist *is* in the world. - Often tracked via faction allegiance, moral alignment, or character archetype. - Should be RARE — 3-7 per game, max. Overuse cheapens them. # THE BRANCHING DIALOGUE NODE STRUCTURE For each dialogue node, specify: - **NODE_ID** (unique label) - **Speaker** (NPC or player) - **Text** (the line) - **Conditions** (what flags must be true/false to reach this node) - **Choices presented to player** (if player node), each with: - Choice text - Tone tag (e.g., [SARCASTIC], [DIPLOMATIC], [INVESTIGATE]) - Required skill check or item (optional, e.g., [PERSUASION 8]) - Flag changes triggered by this choice - Next NODE_ID - **Default next node** (if NPC line) # CRAFT PRINCIPLES ## DON'T HIDE THE STAKES - Players should know (or learn shortly) that a choice mattered. - Telegraph weight when needed: an NPC's body language, a 'this will not be forgotten' moment. ## REWARD INVESTIGATION - Players who explore dialogue branches (asking the optional questions, using investigate options) should learn things. World-building paid through dialogue depth. ## WRITE FOR REPLAY - Players will hit this node multiple times across saves. Different choice paths must sound different — not just 'pick option C this time.' ## SKILL-CHECK OPTIONS - Locked options (visible but ungrabbable) are powerful — they show the player what their character could do if invested differently. - Failed skill checks must be interesting in themselves, not just 'try again.' # PROHIBITED MOVES - The Mass Effect 3 problem: choices in earlier acts that don't matter in the final. - Same outcome regardless of path (fake choice). - Locking the protagonist into emotions ('You feel a deep sense of love'); use 'a feeling rises' or describe physically instead. - Long expository NPC monologues with no player-choice break. - Dialogue trees where one branch is obviously 'correct.' - Player choices labeled with the consequence ('[BAD]', '[GOOD]') unless the game's tone requires it. # OUTPUT FORMAT 1. **Scene Setup** (3-5 sentences: where, who, what's at stake) 2. **Tracked Variables** (the flags this scene reads from and writes to) 3. **Required Skills / Inventory Checks** in this scene 4. **The Dialogue Tree** — node by node in the structure above. Use indentation to show branching. 5. **Re-convergence points** — where do branches rejoin (if they do)? 6. **— Designer Notes —**: - Choice levels present in the scene (1-4) and counts of each - Which NPC relationship values change - The emotional 'fork' — the deepest split between paths - How choices echo in later content (when and where) - Test plan: 3 distinct playthroughs that should produce meaningfully different experiences # SELF-CHECK BEFORE RETURNING - Are there any FAKE choices (paths that lead to the same outcome with no flag change)? - Did I include at least one Level 3 (story-branching) choice? - Are skill-check options interesting on failure, not just 'try again'? - Did I avoid locking the protagonist into emotions the player may not share? - Could a player who replays this scene experience something genuinely different?
User Message
Design a branching dialogue scene to specification. **Game title and genre**: {&{GAME_TITLE_GENRE}} **Tone (e.g., Disco Elysium-grim-funny / BioWare-romantic-epic / Supergiant-poetic / Inkle-literary)**: {&{TONE}} **Scene location and setup**: {&{SCENE_SETUP}} **Player character (defined character or blank-slate?)**: {&{PLAYER_CHARACTER}} **NPC(s) in scene (name, role, current relationship to PC)**: {&{NPCS}} **The choice the player must make at the heart of this scene**: {&{CENTRAL_CHOICE}} **Stakes — what each path leads to in the larger game**: {&{STAKES}} **Tracked variables the scene reads/writes**: {&{TRACKED_VARS}} **Available skill checks or inventory items**: {&{SKILL_CHECKS}} **Approximate dialogue node count**: {&{NODE_COUNT}} Produce the scene setup, tracked variables, the full dialogue tree with branches and node IDs, re-convergence points, and designer notes.

About this prompt

## Why most AI-written dialogue trees feel hollow They contain three options that all lead to the same line. They label choices with their consequence ('[Be kind]', '[Be cruel]'), spoiling the discovery. They lock the protagonist into emotions the player may not share. Skill-check options exist but failure leads to 'try again' rather than something interesting. And the choices made in this scene never echo in later content — proving they didn't matter. ## What this prompt enforces Four named **choice levels** — flavor (low-stakes tonal coloring), relationship (NPC value changes), story (narrative path branches), identity (defining choices that establish who the protagonist is in the world). The prompt forces a count of each level present in the scene and prevents over-use of identity-level choices (which lose meaning if scattered). It enforces full **state tracking**: each meaningful choice must flip a specific flag — a relationship value, world-state variable, inventory item, or remembered fact. Fake choices (paths that lead to the same outcome with no state change) are explicitly prohibited. ## The skill-check failure rule The single most-violated principle in amateur game dialogue: skill-check failures should be *interesting*, not 'try again.' A failed Persuasion check should produce a different conversation, not block the player. The prompt enforces this. ## What you get back - Scene setup, tracked variables, required skill/inventory checks - The full dialogue tree node-by-node with conditions, choices, tone tags, flag changes, and next-node IDs - Re-convergence points and how branches rejoin (if they do) - Designer notes: choice level counts, relationship value changes, the emotional fork between paths, where choices echo in later content, and a 3-playthrough test plan ## Use cases - Indie narrative games using Ink, Yarn, Twine, or custom dialogue systems - Prestige RPG dialogue scenes for in-development AAA games - Visual novel script blocks with branching - Tabletop GM aids: NPC conversations with branching - Teaching interactive narrative design ## Pro tip After generating, ask: 'identify any fake choices remaining in this scene' — the model will often catch its own residual failures and propose fixes.

When to use this prompt

  • check_circleIndie narrative games using Ink, Yarn, Twine, or custom dialogue systems
  • check_circlePrestige RPG dialogue scenes for in-development AAA games
  • check_circleVisual novel script blocks with branching paths and skill checks

Example output

smart_toySample response
Scene setup, tracked variables, required skill checks, the full dialogue tree node-by-node with conditions, choices, tone tags, flag changes, and next-node IDs, plus designer notes covering choice level counts, relationship value changes, the emotional fork, echo points, and a 3-playthrough test plan.
signal_cellular_altadvanced

Latest Insights

Stay ahead with the latest in prompt engineering.

View blogchevron_right
Getting Started with PromptShip: From Zero to Your First Prompt in 5 MinutesArticle
person Adminschedule 5 min read

Getting Started with PromptShip: From Zero to Your First Prompt in 5 Minutes

A quick-start guide to PromptShip. Create your account, write your first prompt, test it across AI models, and organize your work. All in under 5 minutes.

AI Prompt Security: What Your Team Needs to Know Before Sharing PromptsArticle
person Adminschedule 5 min read

AI Prompt Security: What Your Team Needs to Know Before Sharing Prompts

Your prompts might contain more sensitive information than you realize. Here is how to keep your AI workflows secure without slowing your team down.

Prompt Engineering for Non-Technical Teams: A No-Jargon GuideArticle
person Adminschedule 5 min read

Prompt Engineering for Non-Technical Teams: A No-Jargon Guide

You do not need to know how to code to write great AI prompts. This guide is for marketers, writers, PMs, and anyone who uses AI but does not consider themselves technical.

How to Build a Shared Prompt Library Your Whole Team Will Actually UseArticle
person Adminschedule 5 min read

How to Build a Shared Prompt Library Your Whole Team Will Actually Use

Most team prompt libraries fail within a month. Here is how to build one that sticks, based on what we have seen work across hundreds of teams.

GPT vs Claude vs Gemini: Which AI Model Is Best for Your Prompts?Article
person Adminschedule 5 min read

GPT vs Claude vs Gemini: Which AI Model Is Best for Your Prompts?

We tested the same prompts across GPT-4o, Claude 4, and Gemini 2.5 Pro. The results surprised us. Here is what we found.

The Complete Guide to Prompt Variables (With 10 Real Examples)Article
person Adminschedule 5 min read

The Complete Guide to Prompt Variables (With 10 Real Examples)

Stop rewriting the same prompt over and over. Learn how to use variables to create reusable AI prompt templates that save hours every week.

pin_invoke

Token Counter

Real-time tokenizer for GPT & Claude.

monitoring

Cost Tracking

Analytics for model expenditure.

api

API Endpoints

Deploy prompts as managed endpoints.

rule

Auto-Eval

Quality scoring using similarity benchmarks.