How the Brand Identity Pack Template Generates Complete Touchpoint Systems in GPT-Image-2

The brand identity pack template in freestylefly/awesome-gpt-image-2 generates complete touchpoint systems by combining a structured markdown prompt with a reusable JSON schema that directs the image model to render multiple brand applications in a single composite visual board.

The brand identity pack is a core feature of the GPT-Image-2 agent framework that enables one-shot generation of comprehensive visual identity systems. Rather than producing isolated logos or individual assets, this template orchestrates a unified board containing dozens of brand touchpoints—from business cards to storefront signage—while maintaining visual consistency across every panel.

---

Template Architecture in docs/templates.md

The prompt template lives in docs/templates.md under two key sections: "完整品牌身份包模板" and "品牌触点系统视觉板模板" (lines 74-80 and 103-108). This template establishes the structural foundation for touchpoint generation through three layers:

1. Role assignment – The prompt opens with "顶级品牌代理创意总监" (top-tier brand agency creative director), setting the model's perspective and quality expectations.

2. Input parameterization – Required fields include brand name, Chinese name, industry category, target audience, brand keywords, main visual subject, and preferred design style.

3. Explicit deliverable enumeration – The template mandates specific outputs: brand strategy statement, logo concepts with grid construction, color palette definitions, typography system, and—critically—a comprehensive list of application touchpoints.

The touchpoint list in the template specifies concrete deliverables: 名片正反面 (business cards, front and back), 信纸信封 (letterhead and envelopes), APP图标 (app icons), 网站页眉/网站图标 (website headers/favicons), 产品包装/购物袋 (product packaging and shopping bags), and 店面门头/标识牌 (storefront signage).

---

JSON Schema: The brand‑touchpoint‑board Data Model

The data/style-library.json file contains a reusable node at lines 522-539 that codifies this same intent programmatically. The brand-touchpoint-board type serves as the machine-readable bridge between the textual template and the image generation API.

{
  "type": "brand identity system presentation board",
  "header": {
    "title": "品牌视觉识别系统 BRAND IDENTITY SYSTEM",
    "slogan": "爱它·懂它·陪伴它"
  },
  "main_logo": {
    "text": "{argument name=\"brand name\" default=\"GDX\"}",
    "subtitle": "{argument name=\"brand chinese name\" default=\"狗东西\"}",
    "design_feature": "{argument name=\"main subject\" default=\"Dog profile in negative space of the letter D\"}"
  },
  "layout": {
    "sections": [
      { "title": "设计网格", "count": 1, "description": "Logo with architectural grid lines and golden ratio measurements" },
      { "title": "概念草图", "count": 4 },
      { "title": "品牌应用", "count": 6,
        "labels": [
          "名片 正反面",
          "信纸信封",
          "APP图标",
          "网站页眉 / 网站图标",
          "产品包装 / 购物袋",
          "店面门头 / 标识牌"
        ]
      }
    ]
  },
  "constraints": "不只生成一个 logo；保持所有触点统一的配色、排版和图形语言"
}

This JSON structure mirrors the "完整品牌身份包模板" (lines 78-98) and demonstrates how the template defines panel counts, section labels, and cross-touchpoint constraints in a format the agent can parse and inject with user-supplied arguments.

---

Agent Processing Flow

The gpt-image-2-style-library skill orchestrates the generation workflow through three stages:

Argument Injection

The skill reads the JSON template from data/style-library.json, identifies {argument} placeholders, and substitutes user-provided values for brand name, industry, visual subject, and other parameters.

Prompt Construction

The completed prompt combines the role directive, injected parameters, and layout constraints into a single textual instruction. The critical constraint—"所有触点整齐但不死板，主视觉最突出…整体有品牌系统感和可落地感"—instructs the model to arrange all touchpoints in an orderly yet dynamic composition with dominant hero placement and systemic coherence.

Model Interpretation

GPT-Image-2 interprets this as a directive to generate not a single poster but a complete brand application showcase. The model renders:

• A hero product shot as the visual anchor
• Mock-up panels for packaging, tote bags, stickers, menu cards, uniforms, and environmental graphics
• Consistent application of the defined color palette, typography, and graphic motifs across every panel

The agents/skills/gpt-image-2-style-library/references/style-library.md file (lines 314-328) documents this behavior, explicitly linking the brand-touchpoint-board schema to the intended output format of multi-panel brand identity boards.

---

Layout Constraints and Visual Consistency

The template enforces systemic consistency through embedded layout constraints that guide the model's rendering decisions:

Constraint	Purpose
"不只生成一个 logo"	Prevents isolated logo output
"保持所有触点统一的配色、排版和图形语言"	Mandates shared visual system across panels
"主视觉最突出"	Establishes visual hierarchy
"整体有品牌系统感和可落地感"	Ensures practical, deployable appearance

These constraints transform the prompt from a simple asset request into a brand system directive, compelling the model to reason about relationships between touchpoints and maintain coherence across the entire board.

---

Output Characteristics

The generated result is a single composite image containing:

• Strategic header with bilingual brand positioning
• Logo construction panel showing grid systems and proportion rules
• Concept exploration grid with 4 directional variations
• Six-application touchpoint matrix demonstrating real-world deployment
• Unified atmospheric treatment throughout all mock-ups

This output format eliminates the fragmentation typical of iterative logo-then-asset workflows, delivering instead a presentation-ready brand system suitable for client review and production handoff.

---

Summary

• The brand identity pack template in docs/templates.md defines complete touchpoint requirements through parameterized markdown prompts.
• The brand-touchpoint-board JSON schema in data/style-library.json provides reusable, machine-readable structure for agent consumption.
• The gpt-image-2-style-library skill injects user arguments, assembles the final prompt, and dispatches to the image model.
• Layout constraints embedded in the template enforce cross-panel visual consistency and systemic coherence.
• The resulting output is a single composite board demonstrating brand deployment across 10+ touchpoint categories.

---

Frequently Asked Questions

What is a touchpoint system in brand identity design?

A touchpoint system comprises every physical and digital surface where a brand communicates with its audience—business cards, packaging, signage, digital interfaces, and environmental graphics. The brand identity pack template treats these not as separate deliverables but as integrated expressions of a unified visual system, generating them simultaneously in one composite board.

How does the template ensure visual consistency across all generated touchpoints?

The template enforces consistency through explicit constraints in both the markdown template and JSON schema. The directive "保持所有触点统一的配色、排版和图形语言" (maintain unified color, typography, and graphic language across all touchpoints) compels the model to apply the same palette, type system, and visual motifs to every panel, preventing the aesthetic drift common in sequential generation approaches.

Can I customize which touchpoints appear in the generated board?

Yes. The brand-touchpoint-board JSON structure in data/style-library.json exposes the layout.sections array, where you can modify the count and labels fields to add, remove, or rename touchpoint categories. The template's argument system also allows dynamic injection of industry-specific applications (e.g., menu cards for restaurants, app interfaces for SaaS products).

Where is the documentation for extending the brand identity pack template?

Reference documentation resides in agents/skills/gpt-image-2-style-library/references/style-library.md (lines 314-328), which explains the brand-touchpoint-board entry's intent, expected arguments, and output format. The source template in docs/templates.md (lines 74-80, 103-108) provides the authoritative human-readable specification for modification and extension.

{
  "mcpServers": {
    "instagit": {
      "command": "npx",
      "args": ["-y", "instagit@latest"]
    }
  }
}