Skip to content
Skillv1.0.0

image-compress

自动化图片压缩工作流,通过服务端智能压缩实现图片优化。支持本地路径、文件夹和远程 URL,返回 CDN 地址和压缩率。Use when compressing images, reducing file size, or optimizing photos.

by aiskillstore(0) 0 installs
Free
Sign in to install

Free account. Installing gives you the manifest plus copy-paste snippets.

See reviews

About

Imported from aiskillstore/marketplace (skills/xiaowu89/image-compress/SKILL.md). Install upstream with npx skills add aiskillstore/marketplace --skill image-compress. Copyright stays with the author (MIT).

图片压缩

通过 nx-mcp-server 远程压缩服务对图片进行智能压缩。

配置

在项目根目录创建 .mcp.json

{
  "mcpServers": {
    "nx-mcp-compress": {
      "type": "url",
      "url": "https://mcp.api-inference.modelscope.net/4378d43d3e7d4c/mcp",
      "env": {
        "NX_API_KEY": "你的 API Key"
      }
    }
  }
}

查找顺序:项目根目录 → 用户家目录。提取 urlMCP_URLenv.NX_API_KEYAPI_KEY

Skill 直连 MCP 端点,无需重启 Claude Code

没有 API Key? 联系微信 zhjian_2026 获取。

压缩流程

步骤 1:检查配置

cat .mcp.json 2>/dev/null || cat ~/.mcp.json 2>/dev/null
  • 找到 → 记录 urlNX_API_KEY,继续步骤 2
  • 找不到 → 询问用户是否已有 API Key:
    • 有 Key:帮用户创建 ~/.mcp.json,全局和项目安装都通用
    • 没有 Key:告知联系微信 zhjian_2026 获取

⚠️ 配置缺失时不要继续后续步骤。

步骤 2:执行压缩(一次 Bash 调用,纯内存,零文件)

替换 PIC_DIRMCP_URLAPI_KEYQUALITY 后,heredoc 直接通过 stdin 喂给 node:

NODE_PATH=$(npm root -g) node << 'COMPRESSEOF'
const fs=require('fs'),path=require('path');
let PIC_DIR='<目标图片目录绝对路径>';
const SINGLE_FILE='<单张图片路径,为空则压缩整个目录>';
const MCP_URL='<从.mcp.json读取的url>';
const API_KEY='<从.mcp.json读取的NX_API_KEY>';
const QUALITY=90;
(async()=>{
const exts=['.png','.jpg','.jpeg','.webp','.bmp','.tga'];
let imgs;
if(SINGLE_FILE){imgs=[path.basename(SINGLE_FILE)];PIC_DIR=path.dirname(SINGLE_FILE)}
else{imgs=fs.readdirSync(PIC_DIR).filter(f=>exts.includes(path.extname(f).toLowerCase())).sort()}
const origTotal=imgs.reduce((s,f)=>s+fs.statSync(path.join(PIC_DIR,f)).size,0);
console.log(`共 ${imgs.length} 张,总 ${(origTotal/1024).toFixed(0)}KB`);
const maxSize=5*1024*1024;
const records=[];let skip=0;
for(let i=0;i<imgs.length;i++){const f=imgs[i],fp=path.join(PIC_DIR,f),osz=fs.statSync(fp).size;
if(osz>maxSize){records.push({name:f,origKb:osz,compKb:0,compUrl:null,ratio:null,error:'超过5MB限制'});skip++;console.log(`  [${i+1}/${imgs.length}] ${f} ${(osz/1024).toFixed(0)}KB ⚠️ 超过5MB跳过`);continue}
const buf=fs.readFileSync(fp);const ext=path.extname(f).toLowerCase().slice(1);
const mime=ext==='png'?'image/png':ext==='webp'?'image/webp':'image/jpeg';
records.push({name:f,origKb:osz,compKb:0,compUrl:null,ratio:null,dataUrl:'data:'+mime+';base64,'+buf.toString('base64')});}
console.log(`可压缩: ${records.length-skip} 张, 跳过: ${skip} 张`);
console.time('压缩');
const H={'Content-Type':'application/json','Accept':'application/json, text/event-stream'};
const r1=await fetch(MCP_URL,{method:'POST',headers:H,body:JSON.stringify({jsonrpc:'2.0',id:'1',method:'initialize',params:{protocolVersion:'2025-06-18',capabilities:{},clientInfo:{name:'cc',version:'1'}}})});
H['Mcp-Session-Id']=r1.headers.get('Mcp-Session-Id');console.log('MCP: init');
await fetch(MCP_URL,{method:'POST',headers:H,body:JSON.stringify({jsonrpc:'2.0',method:'notifications/initialized'})});console.log('MCP: notified');
// 并发池:始终保持 LIMIT 个请求在飞,完成一个补一个
	const LIMIT=5;
	let totalSaved=0,pass=0,fail=0,done=0;
	const pending=records.filter(r=>!r.error);
	
	async function compressOne(r,idx){
	  try{
	    const r3=await fetch(MCP_URL,{method:'POST',headers:H,body:JSON.stringify({jsonrpc:'2.0',id:'3',method:'tools/call',params:{name:'nx_compress',arguments:{files:[r.dataUrl],quality:QUALITY,apiKey:API_KEY}}})});
	    const raw=await r3.json();
	    if(!raw.result){r.error='MCP: '+(raw.error?.message||'unknown');fail++}
	    else{
	      const inner=JSON.parse(raw.result.content[0].text);
	      if(inner.error){r.error=inner.code+': '+inner.error;fail++}
	      else{const it=inner.items[0];
	        if(it.error){r.error=it.error;fail++}
	        else{r.compKb=it.compressedSize/1024;r.ratio=it.ratio;r.compUrl=it.compressedUrl;totalSaved+=it.originalSize-it.compressedSize;pass++;console.log(`  [${idx+1}/${imgs.length}] ${r.name} ${(r.origKb/1024).toFixed(0)}KB→${(r.compKb).toFixed(0)}KB (${r.ratio})`)}
	      }
	    }
	  }catch(e){r.error=e.message;fail++}
	  done++;
	}
	
	// 滑动窗口:始终保持 LIMIT 个并发
	let i=0;
	const running=new Set();
	while(i<pending.length||running.size>0){
	  while(running.size<LIMIT&&i<pending.length){
	    const p=compressOne(pending[i],i).then(()=>running.delete(p));
	    running.add(p);i++;
	  }
	  if(running.size>0) await Promise.race(running);
	}console.timeEnd('压缩');
console.log('\n'+'='.repeat(100));
console.log(`${'文件'.padEnd(40)} ${'原始'.padStart(8)} ${'压缩后'.padStart(8)} ${'压缩率'.padStart(8)} ${'结果'.padStart(6)}`);
console.log('-'.repeat(100));
for(const r of records){const oszS=(r.origKb/1024).toFixed(1)+'KB';
if(r.compUrl){const cszS=(r.compKb).toFixed(1)+'KB';console.log(`${r.name.padEnd(40)} ${oszS.padStart(8)} ${cszS.padStart(8)} ${(r.ratio||'?').padStart(8)} ${'✅'.padStart(6)} ${r.compUrl}`)}
else{console.log(`${r.name.padEnd(40)} ${oszS.padStart(8)} ${'—'.padStart(8)} ${'—'.padStart(8)} ${'❌'.padStart(6)}`)}}
console.log(`\n📊 ${records.length} 张 | ✅ ${pass} 成功 | ❌ ${fail} 失败 | 共节省 ${(totalSaved/1024).toFixed(0)}KB`);

})();
COMPRESSEOF

压缩参数:quality 默认 90(1-100),无需本地 sharp,直传服务端压缩。

步骤 3:下载到本地(如用户需要)

不要重新运行压缩!压缩结果中的 CDN URL 已在控制台输出,直接用 curl 下载:

curl -o <输出目>/<文件> "<CDN地址>"
  • 成功:压缩后的 CDN URL 可直接使用
  • ⚠️ 超 5MB:文件超过服务端限制,需手动处理
  • 失败:重试一次

返回字段速查

字段 类型 说明
originalSize number 原始大小(bytes)
compressedSize number 压缩后大小(bytes)
ratio string 压缩率,如 "71.5%"
compressedUrl string CDN 地址

常见错误速查

错误现象 原因 正确做法
MISSING_API_KEY 没传 apiKey 必须传
API_AUTH_FAILED API Key 无效 检查 .mcp.json 中的 Key
-32602 Invalid request parameters 未发送 notifications/initialized 必须三步:init → notified → call
406 Not Acceptable 缺少 Accept 同时声明 application/jsontext/event-stream
文件超过 5MB 被跳过 服务端限制 提示用户手动处理

禁止事项

  • ❌ 不要本地压缩(服务端智能压缩,本地 sharp 多余)
  • ❌ 不要传 output 参数(不存在)
  • ❌ 不要省略 apiKey 参数
  • ❌ 不要省略 notifications/initialized 步骤
  • ❌ 不要写任何临时文件(heredoc 直接 stdin,纯内存执行)
  • ❌ 不要把流程拆成多次 Bash 调用(一次 node << 'COMPRESSEOF' 搞定)
  • ❌ 不要使用反斜杠路径,一律用正斜杠(d:/path

Use it

Copy one of these into your project. Installing also returns the manifest and these snippets.

yaml
targets:
  - https://api.opensmartroute.ai/api/v1/registry/aiskillstore-marketplace-image-compress/manifest   # or paste the manifest below

Manifest

An Open Capability Manifest: the router reads it to know what this does, what it costs and when to pick it.

aiskillstore-marketplace-image-compress.ocm.jsonjson
{
  "ocm": "1",
  "id": "aiskillstore-marketplace-image-compress",
  "kind": "skill",
  "name": "image-compress",
  "description": "自动化图片压缩工作流,通过服务端智能压缩实现图片优化。支持本地路径、文件夹和远程 URL,返回 CDN 地址和压缩率。Use when compressing images, reducing file size, or optimizing photos.",
  "publisher": "aiskillstore",
  "version": "1.0.0",
  "capabilities": {
    "domains": [
      "general"
    ],
    "tags": [
      "skill-md",
      "image-compress",
      "compression",
      "optimization",
      "cdn",
      "skills-sh"
    ],
    "languages": [
      "en"
    ]
  },
  "quality_prior": 0.6,
  "examples": [
    "自动化图片压缩工作流,通过服务端智能压缩实现图片优化。支持本地路径、文件夹和远程 URL,返回 CDN 地址和压缩率。Use when compressing images, reducing file size, or optimizing photos."
  ],
  "primary": false,
  "metadata": {
    "source": {
      "provider": "skills.sh",
      "repository": "https://github.com/aiskillstore/marketplace",
      "path": "skills/xiaowu89/image-compress/SKILL.md",
      "ref": "HEAD",
      "url": "https://github.com/aiskillstore/marketplace/blob/HEAD/skills/xiaowu89/image-compress/SKILL.md",
      "key": "aiskillstore/marketplace/skills/xiaowu89/image-compress/SKILL.md"
    },
    "compatibility": "Requires Node.js >= 18 and nx-mcp-server with NX_API_KEY configured",
    "license": "MIT"
  },
  "instructions": "# 图片压缩\n\n通过 nx-mcp-server 远程压缩服务对图片进行智能压缩。\n\n## 配置\n\n在项目根目录创建 `.mcp.json`:\n\n```json\n{\n  \"mcpServers\": {\n    \"nx-mcp-compress\": {\n      \"type\": \"url\",\n      \"url\": \"https://mcp.api-inference.modelscope.net/4378d43d3e7d4c/mcp\",\n      \"env\": {\n        \"NX_API_KEY\": \"你的 API Key\"\n      }\n    }\n  }\n}\n```\n\n查找顺序:项目根目录 → 用户家目录。提取 `url` → `MCP_URL`,`env.NX_API_KEY` → `API_KEY`。\n\nSkill 直连 MCP 端点,**无需重启 Claude Code**。\n\n> **没有 API Key?** 联系微信 `zhjian_2026` 获取。\n\n## 压缩流程\n\n### 步骤 1:检查配置\n\n```bash\ncat .mcp.json 2>/dev/null || cat ~/.mcp.json 2>/dev/null\n```\n\n- 找到 → 记录 `url` 和 `NX_API_KEY`,继续步骤 2\n- 找不到 → 询问用户是否已有 A",
  "cost": {
    "context_tokens": 1482
  }
}

Fetch it by URL: GET /api/v1/registry/aiskillstore-marketplace-image-compress/manifest?version=1.0.0

Reviews

Star ratings from people who tried it. One review per account; edit yours any time.

No reviews yet. Install it, try it, and be the first to rate it.