Hubbestaiskill.com

Home / All Skills / Content & Media Creation / canghe-xhs-images

Content & Media Creation · freestylefly/canghe-skills

canghe-xhs-images

Generates Xiaohongshu (Little Red Book) infographic series with 10 visual styles and 8 layouts. Breaks content into 1-10 cartoon-style images optimized for XHS engagement. Use when user mentions "小红书图片", "XHS images", "RedNote infographics", "小红书种草", or wants social media infographics for Chinese platforms.

L4 High-Privilege · Graded auth & full auditV1 Commit-PinnedExecute CommandWrite FileDual-pass Audit

Author freestyleflyGitHub freestylefly/canghe-skills ↗License MIT（仓库根 LICENSE）Commit dd0bf35595

Control levels describe skill capabilities and suggested control measures; they do not represent quality or safety certifications, nor do they dictate usability. Usage decisions rest entirely with users and administrators. Different agent hosts apply varying constraints; do not assume hosts provide automatic protection. Third-party skill copyrights belong to their original authors. Use at your own risk.

1How It Works & Mechanism

The core is the 2D flexible combination of "Style × Layout": style determines color/lines/decorations, while layout determines information density and arrangement. Both can be arbitrarily paired to generate Xiaohongshu infographics.

skills/canghe-xhs-images/SKILL.md

Style × Layout can be freely combined. Example: `--style notion --layout dense` creates an intellectual-looking knowledge card with high information density.

What this does: Decouples visual presentation into two orthogonal dimensions, allowing the agent to pick both according to content rather than writing art instructions manually per image.

10 visual styles and 8 information layouts are externalized into presets/reference files; SKILL.md only provides an index and one-sentence definitions, reading details on demand.

skills/canghe-xhs-images/SKILL.md

Detailed style definitions: `references/presets/<style>.md`

What this does: Controls context via "main file index + lazy loading references," preventing style details from permanently occupying prompt space.

Features a built-in auto-selection table: recommends styles and layouts based on content signal words (beauty, knowledge, education, study notes, etc.) to reduce user decision-making cost.

skills/canghe-xhs-images/SKILL.md

| Knowledge, concept, productivity, SaaS | `notion` | dense/list |

What this does: Encodes "what style fits this content" into a lookup rule table, enabling the agent to provide suggestions without requiring artistic judgment.

Provides three differentiated outline strategies (story-driven / information-dense / visual-first), producing three structures from the same material for the user to select or merge.

skills/canghe-xhs-images/SKILL.md

Three differentiated outline strategies for different content goals:

What this does: Diverges into three outline drafts before converging, ensuring that "structure" is explicitly selectable just like "style."

The workflow is strictly constrained to "Preference Check (blocking) → Analysis → Confirmation 1 → 3 Outline Drafts → Confirmation 2 → Step-by-step Generation → Report". The initial setup must be completed first if preferences are missing.

skills/canghe-xhs-images/SKILL.md

**CRITICAL**: If EXTEND.md not found, MUST complete first-time setup before ANY other questions or steps. Do NOT proceed to content analysis, do NOT ask about style, do NOT ask about layout — ONLY complete the preferences setup first.

What this does: Defines the initial configuration as a hard gate (MUST) and locks down the state machine with capitalized emphasis to prevent the agent from rushing ahead to analyze.

Two human confirmation checkpoints (content understanding, structure and style/element selection) are marked as REQUIRED to prevent the agent from jumping directly into image generation.

skills/canghe-xhs-images/SKILL.md

**Two confirmation points required** (Steps 2 & 4) - do not skip

What this does: Writes "align before generation" as non-skippable checkpoints, reducing rework and wasted images.

Series consistency relies on a "reference image chain": generate the first cover image without --ref, and use the first image as --ref for every subsequent image to anchor characters and art style.

skills/canghe-xhs-images/SKILL.md

**Use image 1 as `--ref` for ALL remaining images** (2, 3, ..., N)

What this does: Uses the first image as a visual anchor, turning the challenge of "maintaining consistent style across multiple images" into an executable command-appending rule.

Prompts for each image are persisted to disk as prompts/NN-{type}-[slug].md before generation, with a same-name backup rule to ensure changes are reproducible.

skills/canghe-xhs-images/SKILL.md

**Backup rule**: If prompt file exists, rename to `prompts/NN-{type}-[slug]-backup-YYYYMMDD-HHMMSS.md`

What this does: Ensures prompt files exist prior to images, providing an auditable basis for "regenerating/modifying images."

2Core Capabilities

0110 visual styles (cute/fresh/warm/bold/minimal/retro/pop/notion/chalkboard/study-notes)

028 information layouts (sparse/balanced/dense/list/comparison/flow/mindmap/quadrant)

03Automatic style and layout recommendations based on content signals

04Three outline strategies with cross-strategy page merging support

05Interactive preference and element customization (two AskUserQuestion confirmations)

06EXTEND.md preference persistence (watermarks/default style/language)

07Session directory and naming discipline (slug conflict timestamping, source file backups)

08Image modification workflow (edit/add/delete with renumbering)

3External Dependencies

Kind	Dependency / Endpoint
cli	npx / bun（prompt 组装脚本）
package	宿主提供的图像生成 skill（出图依赖，非本目录资产）

4Recommended Governance & Evidence

Suggested Controls for Safe Operation

For Agent Runtime

Graded/per-use authorization, sandbox recommended

For Administrators

Approval workflow, rollback support, full audit

Evidence & Triggers (File Paths & Excerpts)

scripts/: Executes local script or command: references/workflows/prompt-assembly.md instructs to run `npx -y bun ${SKILL_DIR}/script
本地工作目录: File read/write and artifact output: Writes analysis, outlines, prompts, and images to the session directory: `xhs-images/{topic-slug}/` (analysis.md, outline-stra

Audit Notes: Green: All 21 files in this skill directory are Markdown instructions and references, with no bundled executable scripts, no credential reads, and no direct network connections. The only external program reference is in prompt-assembly.md (npx -y bun scripts/main.ts), but this script does not reside in this directory and belongs to the host execution tier, which is not counted toward this skill's tier; image generation relies on the host's image generation skills and external models, and all writes target artifact directories. Rated Green according to unified rating standards.

5Second-Pass Verification

[ok] 外部程序：npx -y bun scripts/main.ts 确实存在 — 扫到 prompt-assembly.md:170 一行；但本 skill 目录清单（§1）无 scripts/ 目录，判定属仓库级资产、非本目录执行。
[ok] 网络请求／恶意外发 — §3 网络请求 0 命中；SKILL.md 与 references 无域名级调用代码。
[ok] 凭证/环境变量读取 — 仅 $HOME 用于定位 EXTEND.md 路径，非凭据泄露。
[ok] 写盘范围 — 写盘 0 命中（正则为线索）；正文明确写入会话目录 xhs-images/{topic-slug}/，属产物。
[ok] 风格/版式枚举与 SKILL.md 声明一致 — §1 列 10 个 presets 文件、4 个 elements 文件、3 个 workflows、3 个 config，与正文引用一致。
[ok] 是否夸大能力 — description 声称 10 风格 8 版式 1-10 图，正文 Style/Layout Gallery 与样式矩阵可对应，未夸大。

6Verdict & Scenarios

Matrix-based flexible combination of styles and layouts covering various Xiaohongshu aesthetics from cute to minimal, offering high reusability.

Strict workflow with two mandatory confirmations and a preference blocking gate to align before rendering, significantly reducing discarded images.

Reference image chain + sessionId + disk-persisted prompts establish clear mechanisms for multi-image consistency and reproducible edits.

Hierarchical lazy loading of references keeps the main file readable and loads details on demand, controlling context window costs.

Best for: Operations teams or content creators who need to batch-produce Xiaohongshu infographics with consistent styles on Chinese social platforms, provided the host already has image generation tools.

Not for: Environments that are offline or lack image generation capabilities; or scenarios requiring strictly reproducible, industrial typesetting (this skill provides artistic stylized generation, not precise layout engine rendering).

Install Agent-Ready

① Verified Mirror 2026-10-10

Option A · Direct DownloadDownload canghe-xhs-images.tar.gz

sha256: d19d5f23ade997a5…

Option B · Copy Guide for Agent

Agents automatically download and verify mirrors from this JSON guide.

② Upstream GitHub

GitHub Repo ↗

Pinned to commit dd0bf35595.