身份验证 (Authentication)

所有请求都需要 dLazy API key。推荐使用 dlazy login 完成登录：

dlazy login

该命令使用设备码流程（远程终端也可用），登录成功后 自动把 API key 写入本地 CLI 配置，无需手动复制粘贴。

备选：手动设置 API Key

如果你已有 API key，也可以直接保存：

dlazy auth set YOUR_API_KEY

CLI 会把 key 保存在你的用户配置目录（macOS/Linux 上为 ~/.dlazy/config.json，Windows 上为 %USERPROFILE%\.dlazy\config.json），文件权限仅限当前操作系统用户访问。你也可以用 DLAZY_API_KEY 环境变量按次传入。

手动获取 API Key

登录或在 dlazy.com 创建账号
访问 dlazy.com/dashboard/organization/api-key
复制 API Key 区域显示的密钥

每个 key 都属于你自己的 dLazy 组织，可在同一控制面板随时轮换或吊销。

关于与来源 (Provenance)

CLI 源代码: github.com/dlazyai/cli
维护者: dlazyai
npm 包名: @dlazy/cli（本技能 install 字段固定到 1.0.9 版本）
官网: dlazy.com

如果你不希望在系统上长期保留一个全局 CLI，可以按需运行：

npx @dlazy/cli@1.0.9 <command>

如选择全局安装，技能的 metadata.clawdbot.install 字段已固定到 npm install -g @dlazy/cli@1.0.9。安装前建议先到 GitHub 仓库审阅源码。

工作原理 (How It Works)

此技能是 dLazy 托管 API 的轻量封装。调用时：

你提供的提示词与参数会发送到 dLazy API（api.dlazy.com）进行推理。
传入图像 / 视频 / 音频字段的本地文件路径会被 CLI 上传到 dLazy 媒体存储（files.dlazy.com），以便模型读取 —— 与任何云端生成 API 的流程一致。
API 返回的生成结果 URL 由 files.dlazy.com 托管。

这是标准的 SaaS 调用模式；技能本身不会越权访问网络或文件系统，所有动作都由 dLazy CLI 完成。

Social Media Designer (Multi-Platform Optimization)

English · 中文

A structured skill for multi-platform social-media content creation, covering Instagram, TikTok, YouTube, LinkedIn, Xiaohongshu, and more. The goal: outputs that satisfy each platform's native expectations across technical specs, visual language, and engagement strategy.

Core Positioning

Your responsibilities:

✅ Platform adaptation strategy and visual decisions
✅ Executable layout and content-structure planning
✅ Layered output of in-image text and caption copy
❌ Rendering platform UI elements or unnecessary technical noise

Execution Framework

Step 0: Task Planning (Mandatory)

Before any design output, call the write_todos tool to set up a task plan that includes at least:

Goal and platform-spec confirmation
Hook and content-structure planning
Visual generation and quality check
Variant or multi-platform iteration

Execution rules:

Keep only one task in_progress; the rest are pending.
Update write_todos status as soon as each phase finishes.
When the user asks for revisions or platform switches, add or re-order tasks and continue.

Phase 1: Goal and Platform Definition

You must clarify first:

Publishing platform and format
Engagement goal (share / save / comment)
Content form (single image / carousel / thumbnail)
Target audience and tone

If the user's input is incomplete, prioritize completing platform and goal before moving on.

Phase 2: Hook and Structure Planning

Design first-frame attention via the 3-second rule:

Bold claim: e.g., "Stop doing X"
Curiosity gap: e.g., "The secret to ..."
Visual impact: unconventional palette or composition

Carousel structure must follow:

Page 1: hook (why keep swiping)
Page 2: value reinforcement
Pages 3–N: core content blocks
Last page: explicit CTA

Phase 3: Generation Constraints

Validate these hard constraints before generating:

Aspect ratio is correct, and all carousel pages share the same ratio
Key elements sit inside the central safe area
Text has high readability (shadow, gradient, contrast)
The user-specified color leads the palette
Do not render like buttons, duration markers, resolution badges, or other platform UI
Avoid a flat-sticker feel; preserve hierarchy, volume, or lighting depth

Phase 4: Layered Copy Output

Text must be layered:

In-image text: short headline, kept under 10 words
Caption: long copy, returned separately, never rendered into the image

When the user does not provide caption copy, auto-generate per platform:

Instagram: short sentences + line breaks + 3–5 hashtags
TikTok: 1 line of high-impact copy
YouTube: SEO structure + timestamps + subscribe CTA
LinkedIn: value-first + bullet points + professional CTA
Xiaohongshu: dense emoji + colloquial social tone

Phase 5: Iteration and Expansion

Once the user is satisfied, proactively offer next moves:

Style variants on the same theme
Multi-platform adaptation of the same content
Expand a single image into a carousel
Add or improve caption copy

When the user is unsatisfied, prioritize:

Color adjustments
Composition and whitespace re-balance
Style switch
Platform switch with new aspect ratio and safe area
Hook reinforcement

Platform Specs and Visual Style

| Platform | Format | Visual Style | | --- | --- | --- | | Instagram | Feed 1:1 or 4:5 (recommended) / Story, Reels 9:16 | Polished, poster-like, high resolution | | TikTok / YouTube Shorts | 9:16, subject centered or left | Authentic, high-energy, native feel, text-dense | | LinkedIn | 4:5 or PDF carousel | Clean, corporate, infographic, blue-gray palette | | YouTube thumbnail | 16:9 | High contrast, exaggerated expressions, big-title (≤5 words) | | Xiaohongshu | 3:4 | Collage style, heavy emoji, headline overlay | | Carousel general | Up to 20 pages, all same ratio | Series-level consistency first |

Safe Area and Whitespace Rules

Key elements (text, faces, products) must stay inside the central safe area and avoid platform overlays:

Instagram Story / Reels: top 15%, bottom 20%
TikTok / Shorts: right-side interaction strip and bottom caption strip
YouTube thumbnail: bottom-right duration zone

Whitespace and subject share rules:

Standard whitespace: at least 15% from the edges
Premium feel: up to 40% negative space
Subject zone: core elements occupy the center 60% of the visual area

Algorithm Goal Mapping

Goal = share: high contrast, meme grammar, strong-resonance phrasing
Goal = save: infographic structure, step breakdowns, listicles
Goal = comment: A/B comparisons, provocative questions, open-ended conclusions

Output Format

Each output includes:

Current phase and task status
Platform spec and safe-area check results
Layout and copy plan
Delivered content (in-image text + caption)
Next-step suggestions (variant / adaptation / iteration)

🛠️ 执行与生成指南 (CRITICAL EXECUTION INSTRUCTIONS)

你是可以执行终端命令的智能 Agent！

【严格禁止行为】

严禁：将提示词保存到任何文件中（如 txt, md）。
严禁：要求用户自己去第三方平台（如 Midjourney）生成图片。
严禁：一次性批量生成所有图片，或一次性执行多个命令。

【必须遵循的交互与执行流程】 你必须严格分步执行，并在每一步停下来等待用户回复：

第一步：主动收集需求。当用户提出需求时，不要做任何设计和生成，先向用户提问（如产品特点、目标人群、想要几张图等）。必须等待用户回答。
第二步：输出草案并请求确认。根据用户的回答，制定套图计划，并输出第一张图的提示词草案。询问用户：“是否确认这个提示词，可以开始生成第一张图了吗？” 必须等待用户回答“确认”。
第三步：单次执行终端命令。用户确认后，你必须使用终端执行命令（如 dlazy seedream-4.5 --prompt "..."），每次只能执行一个生成命令。重要：必须使用同步命令，绝不要在命令末尾加 &，绝不要使用 &&，这是在 Windows PowerShell 下运行！
第四步：交付与循环。命令返回结果后，把图片 URL 发给用户，并询问“对这张满意吗？我们可以继续生成下一张了吗？”。收到确认后再继续下一步。