Create Event

---
name: create-event
description: Interactive questionnaire to create a new game event JSON file for the crawler roguelike. Handles combat, dialogue, skill_check, rest, and boss types. Saves to resources/events/<type>/ and resources/dialogue/ as needed.
model: claude-haiku-4-5-20251001
---

# Create Event

Generate one or more JSON files for a game event. Ask each group, then write the file(s).

## Event types and output files

| Type        | Primary file                              | Secondary file(s)                          |
|-------------|-------------------------------------------|---------------------------------------------|
| combat      | resources/events/combat/<name>.json       | —                                           |
| dialogue    | resources/events/dialogue/<name>.json     | resources/dialogue/<name>.json              |
| skill_check | resources/events/skill_check/<name>.json  | —                                           |
| rest        | resources/events/rest/<name>.json         | —                                           |
| boss        | resources/events/boss/<name>.json         | resources/dialogue/<stem>.json (per trigger)|

## Available enemy scenes

res://scenes/skeleton.tscn res://scenes/skeleton_lord.tscn

---

## Questionnaire

**Q1. Event type** — combat / dialogue / skill_check / rest / boss? [combat]

Then follow the section for the chosen type below.

---

### COMBAT

**Q2.** Event name (snake_case)?
**Q3.** Number of waves? [2]
**Q4.** For each wave: enemy scene (skeleton / skeleton_lord) and count?
**Q5.** Should any wave trigger a dialogue on enemy clear? If yes: which wave (1-indexed) and dialogue file stem? [no]
**Q6.** Rewards — experience and gold? [30, 10]

Output: `resources/events/combat/<name>.json`

```json
{
    "name": "<name>",
    "waves": [
        { "enemies": [{ "scene": "res://scenes/skeleton.tscn", "count": 2 }] },
        { "enemies": [{ "scene": "res://scenes/skeleton.tscn", "count": 3 }] }
    ],
    "rewards": {
        "experience": 30,
        "gold": 10
    }
}

For a wave with dialogue trigger on clear, add "on_clear_trigger": "on_mid" to that wave object.

---

DIALOGUE

Q2. Event name and file stem (snake_case)? Q3. Speaker name(s) — comma-separated, or null for narration? Q4. Opening line (spoken or narration)? Q5. Number of top-level player choices? [2] Q6. For each choice, walk the branch recursively:

• Choice text → speaker response text → further choices or terminal?
• Any rewards on this node (experience, gold)?
• Any consequence (set_flag: value)?

Continue until all paths end in a terminal node (choices: []).

Output two files:

resources/events/dialogue/<name>.json:

{
    "name": "<name>",
    "dialogue": "res://resources/dialogue/<name>.json"
}

resources/dialogue/<name>.json:

{
    "name": "<name>",
    "nodes": {
        "0": {
            "speaker": "Name or null",
            "text": "Opening line.",
            "consequence": null,
            "choices": [
                { "text": "Choice A", "next": "0-0" },
                { "text": "Choice B", "next": "0-1" }
            ]
        },
        "0-0": {
            "speaker": "Name",
            "text": "Response to A.",
            "consequence": { "action": "set_flag", "value": "flag_name" },
            "choices": [],
            "rewards": { "experience": 15, "gold": 3 }
        },
        "0-1": {
            "speaker": null,
            "text": "Narrated response to B.",
            "consequence": null,
            "choices": []
        }
    }
}

Node ID rules:

• Root node: "0"
• Children of "0": "0-0", "0-1", "0-2"
• Children of "0-1": "0-1-0", "0-1-1"
• Terminal nodes have "choices": []
• "consequence" is null or { "action": "set_flag", "value": "<flag>" }
• "rewards" is optional — only include on nodes where rewards are given

---

SKILL_CHECK

Q2. Event name and label shown to the player (e.g. "Sneak past the guard")? Q3. Stat to test — STR / DEF / CON / AGI / SPI / LCK? [AGI] Q4. Rewards on success (experience, gold)? [15, 0] Q5. Rewards on failure? [none] Q6. on_success and on_failure event trigger strings (or empty)? ["", ""]

Output: resources/events/skill_check/<name>.json

{
    "name": "<name>",
    "label": "<label shown to player>",
    "stat": "AGILITY",
    "rewards_on_success": { "experience": 15, "gold": 0 },
    "rewards_on_failure": {},
    "on_success": "",
    "on_failure": ""
}

Stat name strings: STRENGTH, DEFENSE, CONSTITUTION, AGILITY, SPIRIT, LUCK

---

REST

Q2. Event name? Q3. Heal amount — choose a preset or enter a custom expression. Presets: quarter_heal → "max_health * 0.25" half_heal → "max_health * 0.5" spirit_heal → "spirit * 4 + max_health * 0.1" Or enter a custom expression using: health, max_health, strength, defense, constitution, agility, spirit, luck [quarter_heal]

Output: resources/events/rest/<name>.json

{
    "heal_expression": "max_health * 0.25"
}

---

BOSS

Q2. Boss event name? Q3. Number of waves? [4] Q4. Which wave index triggers mid-fight dialogue on clear (1-indexed)? [3] Q5. Enemy scenes per wave (skeleton / skeleton_lord and count)? Q6. Dialogue file stems — intro (on start), mid (on that wave clear), victory (on final clear). Or "none" for any. [none, none, none] Q7. If providing dialogue stems: walk the content of each dialogue file (same branching process as the DIALOGUE type above). Q8. Rewards — experience and gold? [150, 75]

Output: resources/events/boss/<name>.json

{
    "waves": [
        { "enemies": [{ "scene": "res://scenes/skeleton.tscn", "count": 1 }] },
        { "enemies": [{ "scene": "res://scenes/skeleton.tscn", "count": 2 }] },
        {
            "enemies": [{ "scene": "res://scenes/skeleton.tscn", "count": 2 }],
            "on_clear_trigger":