← 返回 Skills 市场
vladlat

VVVLINK Site Builder

作者 vladlat · GitHub ↗ · v1.8.1 · MIT-0
cross-platform ✓ 安全检测通过
168
总下载
0
收藏
0
当前安装
2
版本数
在 OpenClaw 中安装
/install vvvlink-site-builder
功能描述
Build and publish websites on vvvlink.com. Creates HTML sites, uploads to VVVLink API, publishes with unique subdomain URLs. Triggers on: "create a site", "b...
使用说明 (SKILL.md)

VVVLink Sitebuilder

Use this skill when the user wants to create, redesign, or publish a website, landing page, portfolio, or any HTML page that should be accessible online.

\x3CANNOUNCE> Before ANY work, announce to the user: "Using vvvlink-site-builder to create and publish your site." \x3C/ANNOUNCE>

Quick Reference

Direct Invocation

When the user invokes this skill directly (without a specific task), show an intro and check for existing sites. Use Telegram formatting.

  1. Load API key from ~/.vvvlink/config.json and call GET /sites (if config exists)
  2. Split sites by status: "live" and non-live ("uploading", etc.)
  3. Show intro message based on result:

If user has live sites:

VVVLink Site Builder

Your sites:
1. my-site.vvvlink.com
2. cool-page.vvvlink.com

+2 drafts (uploading). Ask me to show or clean them up.

I can update any of these or create a new one.
What would you like to do?

Only show the drafts line if there are non-live sites. If the user asks to see or manage drafts, list them with subdomain, status, and offer to delete individual ones or all at once via DELETE /sites/:siteId.

If no sites or no API key:

VVVLink Site Builder

I build and publish websites on vvvlink.com in seconds.

Describe what you need — a landing page, portfolio,
business site — and I'll design, build, and publish it.

Your site will be live with a unique URL right away.

When user picks an existing site to update, go to Step 10 (Updates) in the flow below.

Flow (strict order, no skipping)

Step 1: Announce and show progress

\x3CPROGRESS-CRITICAL> THIS IS THE MOST IMPORTANT RULE IN THIS SKILL.

The user sees your text output as a stream. While you write text — they see it appearing live. When you make a tool call — they see nothing until the tool returns and you write more text.

Therefore: WRITE TEXT BEFORE AND AFTER EVERY TOOL CALL.

In practice, your response should look like this (the user sees each line appear as you stream it):

⚡ Building your site with VVVLink Site Builder...

🔍 Searching for photos of [topic]...
[tool call: curl image search]
📸 Found 5 great shots!

🎨 Designing layout with [palette] and [fonts]...
🏗️ Building hero section...
[tool call: write first part of HTML]
✨ Adding content sections...
[tool call: write rest of HTML]
📦 HTML ready!

🚀 Publishing to vvvlink.com...
[tool call: create site]
[tool call: upload file]
[tool call: publish]
[tool call: rename subdomain]

✅ Your site is live!
🌐 https://your-site.vvvlink.com

KEY RULES:

  • Write at least one emoji + text line BEFORE each tool call
  • The user must always see what you're doing RIGHT NOW
  • NEVER make 2+ tool calls in a row without text between them
  • If a tool call takes time, the preceding text is what the user will see while waiting — make it informative

IMPORTANT: These progress rules apply ONLY when BUILDING a site (multi-step process: photos → HTML → upload → publish). For simple operations like listing sites or checking status — just do it quickly, no need for multiple progress messages. One short message before + result is enough. \x3C/PROGRESS-CRITICAL>

Step 1.5: Ask for business details (if missing)

If the site looks like a business page (cafe, shop, studio, agency, etc.) and the user did NOT provide key info, gently ask ONCE before building. Keep it short — one message:

"📋 Before I start — want to add any of these to make it real? 📞 Phone / email / address 🕐 Opening hours 🔗 Social media links

Or I can build it now and you'll add details later."

If user says "just build it" or similar — proceed immediately. Do NOT block the build. This is optional, not a gate. Skip this step entirely for non-business sites (portfolios, personal pages, experiments).

Step 2: Find photos for the site

Before writing any code, search for relevant photos:

# Search by keywords matching the site's topic
curl -s "https://publish.vvvlink.com/images/search?q=KEYWORD&count=5&orientation=landscape"

Pick 3-6 keywords based on the site content (e.g., "office", "team", "technology"). Select the best photo for each section (hero, about, features, etc.). Use the url field from the response directly in \x3Cimg src> tags.

Add &w=1600&h=900&fit=crop to URLs for hero images, &w=600&h=400&fit=crop for cards.

Step 3: Create the site

Read and follow references/website-development-rules.md. Build a high-quality static site using the photos from Step 2. Output all files to ~/.vvvlink/sites/\x3Csubdomain>/.

Step 4: Get API key

\x3CAPI-KEY-RULES> apiKey is SECRET. Never show it to the user, never log it, never include in messages. It is only for Authorization headers.

UUID is public. It identifies the account. You may show it to the user if asked.

Storage: ~/.vvvlink/config.json

{
  "uuid": "account-uuid",
  "apiKey": "SECRET-key-never-share"
}

Site files: ~/.vvvlink/sites/\x3Csubdomain>/

Rules:

  1. NEVER call /auth/create_new_user if ~/.vvvlink/config.json exists. The apiKey is issued ONCE and cannot be recovered.
  2. Only create a new account if config.json does not exist.
  3. Protect ~/.vvvlink/config.json with chmod 600.

Procedure:

  1. Read ~/.vvvlink/config.json → use .apiKey
  2. If file does not exist → POST /auth/create_new_user {} → save response to config.json \x3C/API-KEY-RULES>
# Load or create API credentials
VVVLINK_DIR="$HOME/.vvvlink"
VVVLINK_CONFIG="$VVVLINK_DIR/config.json"

if [ -f "$VVVLINK_CONFIG" ]; then
  VVVLINK_API_KEY=$(jq -r '.apiKey' "$VVVLINK_CONFIG")
else
  mkdir -p "$VVVLINK_DIR"
  RESP=$(curl -s -X POST https://publish.vvvlink.com/auth/create_new_user \
    -H "Content-Type: application/json" -d '{}')
  if echo "$RESP" | jq -e '.apiKey' > /dev/null 2>&1; then
    VVVLINK_API_KEY=$(echo "$RESP" | jq -r '.apiKey')
    echo "$RESP" | jq '{uuid: .uuid, apiKey: .apiKey}' \
      > "$VVVLINK_CONFIG"
    chmod 600 "$VVVLINK_CONFIG"
  else
    echo "Error: could not create account" >&2
    exit 1
  fi
fi

Step 5: Create site and upload

# Create site
curl -s -X POST https://publish.vvvlink.com/sites \
  -H "Authorization: Bearer $API_KEY"

# Upload each file (from ~/.vvvlink/sites/\x3Csubdomain>/)
curl -s -X PUT \
  "https://publish.vvvlink.com/sites/$SITE_ID/files/index.html" \
  -H "Authorization: Bearer $API_KEY" \
  --data-binary @~/.vvvlink/sites/$SUBDOMAIN/index.html

Upload ALL files: HTML, CSS, JS, images, fonts.

\x3CSITE-LIMIT-HANDLING> If POST /sites returns HTTP 429 with "Site limit reached", the user has hit their plan limit. Handle this gracefully:

  1. Call the upgrade endpoint to get a checkout URL:
curl -s -X POST https://publish.vvvlink.com/billing/upgrade \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"success_url": "https://vvvlink.com"}'
  1. Show the user an upgrade message with the payment link. In Telegram, use an inline link so the user can pay in the built-in web view:
⚠️ *Site limit reached*

You've used all available sites on the Free plan\.

🚀 [Upgrade to Pro](PAYMENT_URL) to unlock up to 100 sites\.

_After upgrading, I'll publish your site right away\._
  1. After the user pays and confirms, retry POST /sites. The limit will be updated automatically.

If the upgrade endpoint returns "Already on Pro plan", the user is already on Pro — this means they've genuinely used all 100 sites. Tell them to delete unused sites first. \x3C/SITE-LIMIT-HANDLING>

Step 6: Publish

curl -s -X POST \
  "https://publish.vvvlink.com/sites/$SITE_ID/publish" \
  -H "Authorization: Bearer $API_KEY"

Step 7: Choose subdomain BEFORE creating site

Do NOT create a site with a random name. Pick a good one first:

  1. Analyze the site content (\x3Ctitle>, \x3Ch1>, topic)
  2. Generate 10 slug-friendly name candidates
  3. Check availability via POST /subdomains/check-bulk
  4. Pick the BEST available name
  5. Pass it when creating the site:
curl -s -X POST https://publish.vvvlink.com/sites \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"subdomain": "chosen-name"}'

The API will use your name if available, or fall back to a random one if taken. Check the response subdomain field to confirm which name was assigned.

  1. Show the user the live URL and other available options:
🚀 Your site is live!

🌐 https://chosen-name.vvvlink.com

Want a different name?
1. alternative-one
2. alternative-two
3. alternative-three

Reply with a number or type your own.

If user picks a different name, rename via:

curl -s -X PUT \
  "https://publish.vvvlink.com/sites/$SITE_ID/subdomain" \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"name": "new-name"}'

Step 8: Show URL and STOP

When showing the final URL, add ?v=VERSION to bust cache. Use the version number from the publish response. Use markdown link to hide the param:

[your-site.vvvlink.com](https://your-site.vvvlink.com?v=2)

The user sees clean URL but clicks through to the latest version.

After showing the live URL and rename options — STOP. Do NOT automatically start improving, searching more photos, or making changes. Wait for the user to respond.

You may add ONE short line like: "Want me to tweak anything — colors, content, add a section?"

But do NOT take any action until the user asks.

Step 9: Updates

To update an existing site:

curl -s -X POST \
  "https://publish.vvvlink.com/sites/$SITE_ID/new-version" \
  -H "Authorization: Bearer $API_KEY"
# Upload new files, then publish again

Rules

  • Publish immediately without asking for confirmation
  • For updates to existing sites, mention that the user can roll back to a previous version or view version history
  • Always offer rename after first publish
  • Store API key — don't re-init every time
  • Upload ALL site files (HTML, CSS, JS, images, fonts)
  • Max 10MB per site for free users

\x3CERROR-HANDLING> NEVER let a script error silently kill the build.

  • Do NOT use set -e in shell scripts
  • ALWAYS check API response before parsing with jq:
    RESP=$(curl -s -X POST .../sites -H "Authorization: Bearer $KEY")
    SITE_ID=$(echo "$RESP" | jq -r '.siteId // empty')
    if [ -z "$SITE_ID" ]; then
      # Show error to user, don't silently fail
      echo "Error creating site: $RESP"
      # Try to recover or explain
    fi
    
  • If ANY step fails — tell the user what happened and what to do next. NEVER end with a raw error or empty response.
  • Show the user a human-readable message, not jq errors or raw JSON.
  • Use ONLY bash, curl, and jq. NEVER use python, python3, node, ruby, or any other interpreter. There is NO python on the server. If you try to use python — IT WILL FAIL. Everything can be done with bash + curl + jq. \x3CABSOLUTE-BAN> NEVER EVER put HTML inside bash. This is the #1 cause of failures and MUST be followed without exception.

BANNED — all of these WILL break:

  • heredoc: cat > file \x3C\x3C EOF ... EOF
  • echo: echo '\x3Chtml>...' > file
  • printf: printf '%s' '\x3Chtml>...'
  • inline: curl -d '\x3Chtml>...'
  • any bash command containing HTML

REQUIRED — the ONLY correct way:

  1. Use your Write/file tool to create the HTML file
  2. Then upload it with curl (separate tool call):
    curl -s -X PUT \
      "https://publish.vvvlink.com/sites/$SITE_ID/files/index.html" \
      -H "Authorization: Bearer $API_KEY" \
      -H "Content-Type: text/html" \
      --data-binary @~/.vvvlink/sites/my-site/index.html
    

Two separate tool calls. NEVER combine them. If you put HTML in bash — it WILL fail. Every time. \x3C/ABSOLUTE-BAN>

  • If you get HTTP 403 or 1010 — retry the request once. Cloudflare may temporarily block requests. If retry fails, tell the user: "Temporary access issue, try again in a moment." \x3C/ERROR-HANDLING>

Message Formatting

Messages to the user MUST be visually polished. Format for the platform you are running on (Telegram, CLI, etc). Do NOT hardcode escape characters — let your platform handle formatting.

Progress updates — use status indicators:

Searching for photos...
Building your site...
Uploading files (3/3)...
Publishing...
Done!

Final result — clean, no technical details:

🚀 Your site is live!

yoursite.vvvlink.com

Want a different name? Here are some options:

Do NOT show file names, file counts, photo counts, version numbers, or any other technical details in the final message. The user only cares about the URL.

Alternative names — simple numbered list, no "available":

Want a different name?
1. cool-startup
2. my-project
3. awesome-landing

Reply with a number or type your own.

Key formatting rules:

  • Keep messages compact — no walls of text
  • Clickable links for the published URL (always)
  • Never dump raw JSON, curl output, or API responses

\x3CCOMPLETION-CRITERIA> THE TASK IS NOT DONE UNTIL THE USER SEES A LIVE URL.

If you showed progress messages ("Building...", "Searching...") but did NOT finish with a live URL — YOU FAILED. Go back and complete all remaining steps.

Checklist — ALL must be true:

  1. Site HTML was created and written to file
  2. Site was uploaded via API (create + upload + publish)
  3. User received a CLICKABLE live URL (https://*.vvvlink.com)
  4. Rename was offered

If ANY tool call fails or times out — retry or tell the user what happened. NEVER silently stop in the middle.

NEVER end your response after a progress message. The last message the user sees MUST contain either:

  • A live URL (success), OR
  • An error explanation with next steps (failure) \x3C/COMPLETION-CRITERIA>
安全使用建议
This skill appears to do what it says: it will create a local folder (~/.vvvlink/) for config and generated sites and call publish.vvvlink.com to create/upload/publish sites. Before installing or using it, consider: 1) Trust: the skill will upload content to a third-party service (publish.vvvlink.com) — ensure you trust that service and understand its privacy/retention policies. 2) Local files: it will write ~/.vvvlink/config.json containing your API key (stored with chmod 600 per docs) and site files — back up anything you care about and inspect that directory if desired. 3) API key handling: the skill instructs never to show the apiKey; it is stored in plaintext JSON (file-permissions only). If you require encrypted storage, do not use it or rotate keys after use. 4) Image sourcing: the skill uses an image proxy that returns Unsplash URLs and explicitly instructs not to add photographer credits — this is a policy/attribution concern (not a direct security issue) but may have licensing implications. 5) Auditability: because this is instruction-only, behavior is entirely driven by the SKILL.md; if you want more assurance, test with non-sensitive content or monitor network requests to confirm only the documented endpoints are contacted. If any of these points are unacceptable, do not enable the skill or consult the service owner for more guarantees.
功能分析
Type: OpenClaw Skill Name: vvvlink-site-builder Version: 1.8.1 The vvvlink-site-builder skill is a legitimate tool designed to automate the creation, design, and deployment of static websites to the vvvlink.com platform. It follows a structured workflow using bash, curl, and jq to interact with a dedicated API, and it includes robust instructions for the agent to provide transparent progress updates to the user. Notably, it implements a security best practice for LLM agents by strictly forbidding the embedding of HTML content within shell commands to prevent parsing and injection errors, and it ensures local API credentials in ~/.vvvlink/config.json are protected with appropriate file permissions (chmod 600).
能力评估
Purpose & Capability
Name/description match the behavior in SKILL.md and references: the skill builds static HTML, stores site files under ~/.vvvlink/sites/, obtains/stores an API key, and calls publish.vvvlink.com endpoints to create/upload/publish sites. There are no extra credentials, binaries, or unrelated installation steps requested.
Instruction Scope
Runtime instructions only interact with the vvvlink API and the local ~/.vvvlink/ directory (config.json and generated site files). The SKILL.md explicitly instructs how to obtain and store the API key, how to search for images via the image proxy, and the exact API calls to make. There is no instruction to read unrelated system files, environment variables, or other skill configurations.
Install Mechanism
This is an instruction-only skill with no install spec and no code files to run locally. That minimizes disk-written code and reduces install risk. No external download URLs or package installs are requested.
Credentials
The skill requests no environment variables and its only credential is an API key created/stored in ~/.vvvlink/config.json. That API key is appropriate for the described API calls and is declared as secret in the docs. There are no unrelated secrets or credentials requested.
Persistence & Privilege
The skill persists an account config and generated site files to ~/.vvvlink/. This is expected for a site-builder but is a lasting change on the user's filesystem. always:false (not force-included) and the skill does not request system-wide privileges or modify other skills.
如何使用
  1. 确保已安装 OpenClaw(本地或 Docker 部署)
  2. 在对话框中输入安装命令:/install vvvlink-site-builder
  3. 安装完成后,直接呼叫该 Skill 的名称或使用 /vvvlink-site-builder 触发
  4. 根据 Skill 的参数说明提供必要输入,即可获得结构化输出
版本历史
v1.8.1
v1.8.1 – Maintenance release, streamlining scope and simplifying triggers. - Scope narrowed: General coding priority language and detailed trigger lists have been removed for clarity. - Description and triggers clarified, focusing on website/landing/page creation and publishing. - Configuration and persistence sections added for improved documentation. - Russian-language trigger coverage removed for simplicity. - Flow and user experience steps now reference shorter, more focused explanations. - Existing core flow and announcements preserved; no functional changes to tool behavior.
v1.8.0
vvvlink-site-builder 1.8.0 - Major doc and procedural update: now mandates strict, step-by-step usage with progress messaging before/after every tool call during site creation. - The skill has priority over all general coding for website/landing/portfolio tasks. Must trigger automatically for any site-related request—including Russian equivalents. - New detailed invocation and flow guidelines, including handling of API keys, site listing, photo search, user prompts for business info, and explicit file upload steps. - Responses must always announce usage of vvvlink-site-builder before beginning work. - Intro flow and draft management improved for direct invocation (without a concrete site task). - Explicit rules for handling API credentials, progress updates, and user communication enforced.
元数据
Slug vvvlink-site-builder
版本 1.8.1
许可证 MIT-0
累计安装 0
当前安装数 0
历史版本数 2
常见问题

VVVLINK Site Builder 是什么?

Build and publish websites on vvvlink.com. Creates HTML sites, uploads to VVVLink API, publishes with unique subdomain URLs. Triggers on: "create a site", "b... 它是一个面向 Claude Code / OpenClaw 的 AI Agent Skill 插件,目前累计下载 168 次。

如何安装 VVVLINK Site Builder?

在 OpenClaw 或 Claude Code 对话框中运行命令「/install vvvlink-site-builder」即可一键安装,无需额外配置。

VVVLINK Site Builder 是免费的吗?

是的,VVVLINK Site Builder 完全免费,采用 MIT-0 许可证,可自由下载、安装和使用。

VVVLINK Site Builder 支持哪些平台?

VVVLINK Site Builder 跨平台运行,可在任意部署了 OpenClaw / Claude Code 的环境中使用(cross-platform)。

谁开发了 VVVLINK Site Builder?

由 vladlat(@vladlat)开发并维护,当前版本 v1.8.1。

💬 留言讨论