Brand Gene Style Extraction (Brand Gene Extract)

Based on product images and user-provided brand gene parameters, extract a unified brand visual language (Brand DNA) and output structured JSON for consumption by downstream atomic skills.

Applicable Scenarios

Scenario	Description
Image Layout Call	Called by `linkfox-aigc-imagegen-cloth` / `linkfox-aigc-imagegen-product` (image layout path) in Step 3 to define a unified visual gene for the entire set of images
Brand Gene Change	Re-extract after users modify main color/font/platform/region
Initial Brand Definition	First extraction when no brand gene results exist in the context

Inapplicable Scenarios

When a complete brandGeneJson already exists and no changes are needed (reuse directly, do not call this skill repeatedly)
Pure background replacement/image matting/synthesis and other image editing operations
Non-visual brand story copywriting

Input Parameters

Parameter	Type	Default	Description
images	list[image]	—	List of product images uploaded by the user (at least 1 image), used to analyze product colors and temperament
brandKey	object	—	Brand gene information entered by the user (see field description below)

brandKey Field Description

Field	Type	Default	Description
brandColor	string	""	Brand main color (HEX value), use directly if non-empty; auto-extract if empty
fontStyle	string	""	Font style, use directly if non-empty; auto-extract if empty
brandName	string	""	Brand name, used for brand implantation strategy
language	string	"English"	Target market language
platform	string	"Amazon"	Publishing platform (Amazon/TikTok/Shopee, etc.)
salesRegion	string	"United States"	Sales region/country

Pipeline Steps

Step 1: Parameter Parsing and Decision Routing

Input:
```
brandKey
```
,
```
images
```
Operations:
1. Parse the values of each field in
```
brandKey
```
2. Main Color Decision: If
```
brandColor
```
  is non-empty → use directly; if empty → mark as "auto-extract"
3. Font Decision: If
```
fontStyle
```
  is non-empty → use directly; if empty → mark as "auto-extract"
4. Fill Default Values: If
```
language
```
  is empty, default to "English"; if
```
platform
```
  is empty, default to "Amazon"; if
```
salesRegion
```
  is empty, default to "United States"
5. Convert
```
images
```
  to a list of accessible URLs (
```
imageUrlList
```
  ) for passing to textgen in Step 3
Output: Decision routing table (which fields are provided by the user, which need auto-extraction) +
```
imageUrlList
```
Purpose: Guide prompt assembly in Step 2

Step 2: Assemble Brand Visual Gene Extraction Prompt

Input: Decision routing table from Step 1, all fields of
```
brandKey
```
Operations: Assemble the following brand visual gene extraction instructions with
```
brandKey
```
parameters into a complete prompt, and pass it to Step 3:
Role: You are a top brand visual expert and creative director. Based on product images and parameters, build a set of unified visual language to ensure high brand consistency across different scenarios.

A. UNIFIED_VISUAL_THEME
1. Brand Color:
  - Define a core color with a "worldview", not just a color palette
  - Uniqueness: Only 1 core HEX color value can be defined for the entire project
  - Source: Must consider the product's own color, with extremely high aesthetics
  - ⚠️ Black & White & Gray Protocol (CRITICAL): Unless the product itself is black, white, or gray, it is forbidden to use black, white, or gray as the main color. A specific color value (e.g., #EAF86C) must be output
2. Background Strategy: Must fully output the following 4 sub-fields, no exceptions, each with specific content (no empty fields, no field names only):
  - Background Strategy - Style Definition: Create a culturally localized overall environmental style based on
    salesRegion
    . Pursue 100% photo-realistic environment, strictly prohibit layered background texture, focus on lifestyle and emotional resonance. Example:
    "Modern Nordic minimalist home, log and warm white tones, natural lighting through floor-to-ceiling windows, in line with American middle-class aesthetic"
  - Background Strategy - Scene Keywords: Specific scene/prop keywords separated by commas. Example:
    "log long table, linen tablecloth, green plants, ceramic utensils, morning light"
  - Background Strategy - Light & Shadow: Clear light direction, color temperature and atmosphere (this is a required field, most likely to be missed, must output). Example:
    "Soft natural side light, warm color temperature 4000K, creates a lazy morning atmosphere, slight long projection enhances three-dimensionality"
  - Brand Injection: How brand main color / Logo / logo elements are naturally integrated into the scene (this is a required field, most likely to be missed, must output). Example:
    "Brand main color appears as an accent color on pillows/tags, Logo appears on props in a low-saturation embossed form, not overwhelming the main subject"
B. LOCKED_TYPOGRAPHY
1. Font Style: Lock to only one style from the following library:
  - Geometric Sans-serif: Jost, League Spartan, Montserrat, Manrope, Outfit
  - Bold Sans-serif: Bebas Neue, Oswald, Barlow Condensed, Anton, Fjalla One
  - Classic Elegant Serif: Bodoni Moda, Playfair Display, Prata, Cormorant Garamond, DM Serif Display
  - Rounded Playful Fonts: Nunito, Quicksand, Varela Round, Fredoka, Baloo 2
  - Playful Handwritten Style: Pacifico, Permanent Marker, Amatic SC, Caveat, Luckiest Guy
2. Color Strategy:
  - Heading Color: Must use brand main color or reverse white; automatically trigger "flexible reverse white" when the color at the bottom of the heading is brand main color or same color system; secondly consider using colors in the same color system as brand main color (prohibit colors close to black, colors cannot be too dark)
  - Body Color: Distinguish from heading color to ensure readability
  - ⚠️ Output Format Note: The color strategy output format is
    ["Heading Color": color value]
    and
    ["Body color": color value]
    , this is color definition rather than text content
3. Flexible Reverse White Authorization: Authorize switching to matte white (#FFFFFF) text when using dark background or solid brand-color panel
4. Typesetting Restrictions: Unified as non-italic, standard leading
Output Requirements: Strictly output complete JSON according to the "Output Format" below, no explanatory text or code block markers are allowed.
Output: Complete textgen prompt string
Purpose: As prompt input for Step 3

Step 3: Call linkfox-aigc-textgen to Perform Visual Reasoning

Input: Prompt from Step 2,
```
imageUrlList
```
from Step 1
Operations: Call textgen to perform visual reasoning according to the calling method in linkfox-aigc-textgen SKILL.md. The content produced in this step should be read back into the context for field extraction (Step 4), not chained into downstream image generation. Therefore, use textgen's default mode (not
```
--content-only
```
): Write the following fields into a JSON parameter file, pass it in via
```
--stdin
```
to run, then parse stdout.
```
<textgen root directory>
```
is obtained by parsing the absolute path of the directory where SKILL.md is located via
```
skill:linkfox-aigc-textgen
```
;
```
<this skill root directory>
```
= absolute path of the directory where this SKILL.md is located.
⚠️ Intermediate parameter files are stored in the session
data/
directory. First get the directory:
bash
```
DATADIR=$(python <this skill root directory>/scripts/save_brand_gene.py --datadir)
```
⚠️ Construct
brand_gene_params.json
must use the Write tool to write to
$DATADIR/brand_gene_params.json
(prompt contains special characters such as quotes and backslashes, shell heredoc embedded JSON is prone to parsing failures). If generating via command line, only use
python -c
with
json.dumps
, do not manually拼 JSON strings.
bash
```
python <textgen root directory>/scripts/aigc_textgen.py --stdin < "$DATADIR/brand_gene_params.json"
```
The JSON parameter file contains the following fields:
- ```
prompt
```
  : Brand visual gene extraction instructions assembled in Step 2 (including "Output Format" requirements)
- ```
imageUrls
```
  :
```
imageUrlList
```
  output from Step 1
- ```
model
```
  :
```
GEM_3_1_PRO
```
  (requires visual understanding and complex reasoning capabilities)
- ```
thinkingLevel
```
  :
```
high
```
Output: Parse textgen stdout to get
```
content
```
(parsing method follows the "Output Contract" in linkfox-aigc-textgen SKILL.md).
Purpose: As assembly material for Step 4

Step 4: Assemble brandGeneJson

Input:
```
content
```
parsed from Step 3 (line breaks may be
```
⏎
```
placeholders, restore to line breaks before extraction)
Operations: Extract each field of brand visual gene from the
```
content
```
, and assemble it into
```
brandGeneJson
```
strictly according to the "Output Format" below
Missing Field Fallback (Mandatory): After assembly, must check one by one whether all fields in the "Output Format" are complete. If the model output misses a certain field (the most likely to miss
```
Background Strategy - Light & Shadow
```
and
```
Brand Injection
```
), do not directly deliver incomplete JSON, and must fallback and fill in according to the following order:
1. Prioritize inferring a reasonable value for the field from other text in
```
content
```
  ;
2. If there is no inferable content, generate a fallback description consistent with the overall tone based on the determined main color / style definition (e.g., light & shadow as "natural soft side light, warm color temperature, enhanced three-dimensionality", brand injection as "main color as accent, Logo integrated into props in a low-key manner"). The final
```
brandGeneJson
```
  must have complete fields, no missing fields or empty field values are allowed.
Output:
```
brandGeneJson
```
(JSON list with length 1, complete fields)
Storage (Mandatory): After assembly, must write to the session directory and retain the path for downstream reuse:
1. Use the Write tool to write
```
brandGeneJson
```
  to
```
$DATADIR/brand_gene_assembled.json
```
  (
```
DATADIR
```
  same as Step 3); or
2. Run the storage script (recommended, automatically registers
```
_meta.json
```
  ):
  bash
```
python <this skill root directory>/scripts/save_brand_gene.py "$DATADIR/brand_gene_assembled.json"
```
  stdout will print
```
Saved full response: <absolute path> (<N> bytes)
```
  ——Record this absolute path into the context, and subsequent S3 image layout consumes it via
```
--brand-gene-file <absolute path>
```
  or manifest's
```
brand_gene_json
```
  .
Purpose: Pass to downstream image generation links (such as种草图/selling point image/A+ image of
```
linkfox-aigc-imagegen-cloth
```
/
```
linkfox-aigc-imagegen-product
```
, etc.) as
```
brandGeneJson
```
parameter

Product Storage

Product	Directory	Description
`brand_gene_params.json`	`<session>/data/`	textgen input parameters (intermediate file in Step 3)
Final `brandGeneJson` result	`<session>/data/linkfox-aigc-imagegen-brand-gene-extract-<ts>.json`	Stored via `save_brand_gene.py` , reused by S3 image layout and single image `--brand-gene-file`

Path Protocol:

<cwd>/linkfox/<YYYY-MM-DD>/<session>/data/

(

<session>

is obtained from

SESSION_ID

), consistent with "Output File Location" in warehouse

CLAUDE.md

Core Rules

A. Main Color and Font Logic

If
```
brandColor
```
has a value → directly use this value for "brandColor (Brand Main Color)" in the output JSON
If
```
fontStyle
```
has a value → directly use this value for "Font Style" in the output JSON
If any value is empty → extract this value automatically using the rules in core variables

B. Other Value Logic

Fields such as "Background Strategy - Style Definition" and "Color Strategy - Heading" are obtained using the rules in core variables by default, and are not affected by the passing of brandColor/fontStyle in brandKey.

Output Format

Strictly output according to the following JSON format, output as a list with length 1:

json

[
  {
    "brandColor": {
      "brandColor (Brand Main Color)": "Color name and HEX value",
      "Background Strategy - Style Definition": "Style description based on salesRegion",
      "Background Strategy - Scene Keywords": "Keyword1, Keyword2, ...",
      "Background Strategy - Light & Shadow": "Light and shadow atmosphere description",
      "Brand Injection": "How brand elements are implanted into the scene"
    },
    "fontStyle": {
      "Font Strategy": "Selected font family name",
      "Font Style": "Selected specific font name",
      "Color Strategy - Heading": "[\"Heading Color\": color value/reverse white logic]",
      "Color Strategy - Body/Sub": "[\"Body color\": color value]",
      "Flexible Reverse White": "You are authorized to switch to Matte White (#FFFFFF) text whenever using a dark background or a solid brand-color panel.",
      "Typesetting": "Non-italic, standard leading"
    }
  }
]

Error Handling

Situation	Handling
images is empty	Stop execution, prompt the user to upload at least 1 product image
brandKey is completely empty	Fill all fields with default values (salesRegion=United States, language=English, platform=Amazon), main color and font are auto-extracted
Product is pure black, white, or gray and no color main color can be extracted	Allow using black, white, or gray as the main color (the only exemption condition for the black & white & gray protocol)

Calling Method

This skill is called by the image layout layer in Step 3, calling timing:

First Call: When brand gene style extraction has not been performed in the context
Re-call: When the user explicitly requests to modify the brand gene (Scenario G)

Reuse Rule: Extraction results in the same context can be reused, no need to call repeatedly.

Resolve Authentication and Points Issues

When the following abnormal situations occur, guide users to solve problems using references/onboarding.md:

Abnormal Situations

API Key Not Configured: Environment variable
```
LINKFOX_AGENT_API_KEY
```
is not configured, nor is
```
LINKFOXAGENT_API_KEY
```
configured.
Response with 401 or 402 Status Code
Response Indicates Insufficient Points or Balance: Message contains "insufficient points balance/insufficient billing/insufficient balance/quota exceeded/insufficient balance/package expired/need to recharge/please recharge", or similar meanings.

Known Limitations

Brand main color extraction is based on
```
GEM_3_1_PRO
```
's visual understanding of product images, which may deviate from professional designers' judgments
Font styles are limited to 25 fonts in 5 preset categories, cannot select fonts outside the library
Cultural localization of background strategy depends on semantic understanding of salesRegion, no structured regional database
Brand gene JSON must be stored in the session
```
data/
```
via
```
save_brand_gene.py
```
, cannot only be stored in the conversation context
Depends on the availability of
```
linkfox-aigc-textgen
```
skill; if textgen call fails, brand gene extraction cannot be completed

linkfox-aigc-imagegen-brand-gene-extract

NPX Install

Tags

SKILL.md Content (Chinese)