Cleanor

app.cleanor由 cleanor.app 發布官方發布者v0.6.1更新於 Oct 24, 2026

Zero-auth MCP: image optimize, cited storage/format data, and dev utilities LLMs get wrong.

已驗證Streamable HTTP可網頁執行開發者工具資料與分析媒體與設計

概覽

AI 產生的概覽

免驗證的 MCP 伺服器,提供 22 個工具:圖片最佳化、QR 碼與佔位圖產生、附引用的儲存與格式資料,以及確定性的開發小工具。

可以做什麼
  • 從公開網址抓取圖片並重新編碼成更小的 WebP、AVIF 或 JPEG,可選擇依目標寬度縮放,並產生 SVG QR 碼與佔位圖
  • 查詢 Cleanor Labs 研究中附引用的儲存容量與圖片格式節省數據
  • 對文字計算 SHA 摘要與 HMAC 簽章,並解碼 JWT(不驗證簽章)
  • 產生 UUID、Base64 編解碼、JSON 驗證與格式化、正規表達式測試、兩段文字逐行比對
  • 單位、進位、顏色與配色方案轉換,解析網址,解釋 cron 表達式,精確計數以及依 IANA 時區回傳目前時間
適用情境
適合需要精確、確定性結果的情境,例如雜湊、計數、進位換算、時區、正規表達式驗證等模型容易算錯的操作;也適合壓縮圖片、產生 QR 碼或佔位圖,或引用實測的儲存與格式節省數據。
執行需求
使用託管端點無需安裝,任何支援 Streamable HTTP 的 MCP 用戶端指向伺服器網址即可。無需帳號、API 金鑰或 OAuth。可選的本機 stdio 版本透過 npx 執行,圖片編碼依賴 sharp,其餘 21 個工具無需額外依賴。託管端點需要網路存取,optimize_image 還會抓取你提供的圖片網址。
安裝前請注意
託管端點依 IP 限流(一般每分鐘 120 次,圖片最佳化每分鐘 30 次)。輸入內容會傳送到託管伺服器;optimize_image 還會抓取你提供的公開圖片網址,敏感資料建議改用本機 stdio 版本。hmac 需要以工具參數傳入密鑰,jwt_decode 不驗證簽章,解出的宣告不能視為已驗證內容。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 Cleanor,將其新增到工作區。
  2. 為需要使用其工具的對話啟用該服務。

Web executable,透過 Streamable HTTP。 遠端服務在工作區中設定後即可從網頁執行環境執行。

交給代理程式安裝

把這段提示詞貼上到 Claude Code、Codex、Cursor 等代理程式,或 SourceWeft 對話中。代理程式會讀取該伺服器的安裝指南,先向你展示發布者、端點和登入方式,經你同意後再把它新增到你的客戶端。金鑰不會寫進任何命令。

閱讀 https://sourceweft.com/mcp/app.cleanor%2Fcleanor/install.md,並按其中說明把這個 MCP 伺服器接入我的客戶端。修改任何設定之前,先告訴我它的發布者、端點和登入方式,等我確認。不要把 API 金鑰寫進命令或我會提交的檔案;問我金鑰應從哪裡取得。

檢視代理程式所依據的安裝指南

其他 MCP 客戶端

把它新增到你客戶端的 mcpServers 設定中。

{
  "mcpServers": {
    "cleanor": {
      "type": "http",
      "url": "https://mcp.cleanor.app/mcp"
    }
  }
}

README

Cleanor MCP: a zero-auth MCP server for image optimization and dev utilities

A hosted, zero-auth MCP server with 22 tools: optimize and convert images, generate QR codes, and run the deterministic dev utilities LLMs get wrong.

[npm] [license] [MCP Registry] [smithery] [DOI]

No API key. No signup. No OAuth. Point any Model Context Protocol client at one URL and 22 tools appear, all read-only and safe to hand to an autonomous agent.

Streamable HTTP endpointhttps://mcp.cleanor.app/mcp
npm package (stdio)@cleanor/mcp
Official MCP Registryapp.cleanor/cleanor
Homepagecleanor.app/mcp
Authnone
Tools22
Version0.6.0

Install

Pick your client. Every option below points at the same hosted server, so there is nothing to build, install or key.

Claude Code

bash
claude mcp add --transport http cleanor https://mcp.cleanor.app/mcp

Then run claude mcp list to confirm it connected. Full walkthrough: docs/add-an-mcp-server-to-claude-code.md.

Cursor

Add to .cursor/mcp.json (per project) or ~/.cursor/mcp.json (global):

json
{  "mcpServers": {    "cleanor": {      "url": "https://mcp.cleanor.app/mcp"    }  }}

VS Code

Run code --add-mcp '{"name":"cleanor","type":"http","url":"https://mcp.cleanor.app/mcp"}', or add the server to .vscode/mcp.json in the workspace:

json
{  "servers": {    "cleanor": {      "type": "http",      "url": "https://mcp.cleanor.app/mcp"    }  }}

Claude Desktop

Open Settings, then Connectors, then "Add custom connector", and paste the endpoint:

https://mcp.cleanor.app/mcp

Or run the server locally over stdio from claude_desktop_config.json:

json
{  "mcpServers": {    "cleanor": {      "command": "npx",      "args": ["-y", "@cleanor/mcp"]    }  }}

The npm build is the same tool registry compiled for Node. It uses sharp (an optional dependency) for local image encoding; the other 21 tools need nothing.

Raw Streamable HTTP

The endpoint speaks the MCP Streamable HTTP transport, so you can drive it with curl or any HTTP client:

bash
curl -s https://mcp.cleanor.app/mcp \  -H 'content-type: application/json' \  -H 'accept: application/json, text/event-stream' \  -d '{"jsonrpc":"2.0","id":1,"method":"initialize",       "params":{"protocolVersion":"2025-06-18","capabilities":{},                 "clientInfo":{"name":"curl","version":"1"}}}'

There is also a plain REST surface for non-MCP callers: GET /v1/capabilities (feature detection, machine-readable tool list) and POST /v1/optimize (one image in, optimized bytes out).

Tools

22 tools, every one of them carrying an input and an output schema, so a client gets typed structuredContent back and not just a wall of text.

Image

ToolWhat it doesKey params
optimize_imageFetches an image from a public URL and re-encodes it smaller. Returns the optimized image plus before and after byte counts. Input capped at 25 MB.image_url (required), format (webp | avif | jpeg, default webp), width (16 to 4096 px, optional), quality (1 to 100, default 80)
placeholder_imageDependency-free SVG placeholder at any size, with a label and custom colors.width (default 600), height (default 400), bg, color, text
qr_codeEncodes text or a URL as a crisp SVG QR code you can paste anywhere.text (required, up to 2000 chars), ecc (L | M | Q | H, default M), size (64 to 1024, default 320)

Cited data (Cleanor Labs research)

ToolWhat it doesKey params
storage_capacityHow many photos or minutes of video actually fit in a storage tier, corrected for real OS and filesystem overhead. Returns a per-format breakdown.storage_gb (required, 1 to 4096), content (photos | video, default photos)
image_format_savingsHow much smaller WebP, AVIF or JPEG XL are than JPEG at matched perceptual quality, from a controlled benchmark on the 24-image Kodak suite. Also reports the HEIC conversion tax.format (webp | avif | jxl, default avif), quality (web = SSIM 0.95, high = SSIM 0.98, default web)

Dev utilities

The operations an LLM is unreliable at: hashing, counting, radix math, timezones. Pure compute, no network, no state.

ToolWhat it doesKey params
hashSHA-1/256/384/512 hex digest of text. MD5 is deliberately not offered.input (required), algorithm (default sha-256)
hmacKeyed HMAC signature, hex or Base64. For signing and verifying webhooks.message (required), secret (required), algorithm (default sha-256), encoding (hex | base64)
uuidUUID v4 (random) or v7 (time-sortable, the better database key).version (v4 | v7, default v4), count (1 to 100)
base64Base64 encode or decode, UTF-8 safe, optional URL-safe alphabet.input (required), mode (encode | decode), url_safe (boolean)
json_formatValidates JSON, then pretty-prints or minifies it, optionally deep-sorting keys. Returns a precise parse error if invalid.input (required), mode (pretty | minify), sort_keys (boolean)
jwt_decodeDecodes a JWT header and payload and flags expiry. The signature is not verified and no secret is needed.token (required)
regex_testRuns a JavaScript regex against sample text and returns every match with its index and captured groups (first 100).pattern (required), input (required), flags (default g)
cron_describeExplains a 5-field cron expression in plain English and lists the next 5 run times in UTC.expression (required)
unit_convertExact conversion across length, mass, data size, time, speed and temperature.value (required), from (required), to (required)
datetimeThe real current time, or any timestamp, rendered in an IANA timezone as ISO, Unix and human forms.input (optional Unix timestamp or ISO string), timezone (default UTC)
url_parseSplits a URL into scheme, host, port, path, decoded query params and fragment.url (required)
base_convertInteger conversion between bases 2 and 36, BigInt-exact so large values stay exact.value (required), from_base (default 10), to_base (default 16)
diffLine-by-line diff of two texts with an added and removed count. Up to 1000 lines per side.a (required), b (required)
colorConverts one color into hex, RGB and HSL at once. Accepts hex, rgb(), hsl() and CSS names.value (required)
color_paletteDerives a harmonious palette from one base color.color (required), harmony (complementary | analogous | triadic | tetradic | monochromatic)
slugifyClean, URL-safe slug from a title: lowercased, hyphenated, accents stripped.input (required), separator (- | _)
countExact characters (code points), UTF-16 units, words, lines and UTF-8 bytes.input (required)

Every tool is annotated readOnlyHint: true. Only optimize_image touches the network (openWorldHint: true); the other 21 are pure functions of their input.

Full input and output schemas: docs/mcp-tools-reference.md.

Tool index (CSV)

data/tools.csv is the same 22 tools as a flat, machine-readable table: name, title, category (image, data, dev), summary, required and optional parameters, and output fields. Handy for diffing releases, feeding an MCP directory, or comparing this server against another.

The file is generated, not hand-written: every value is read out of the tool's Zod schema in src/tools/. If you add or change a tool, regenerate it.

bash
npm run tools:csv        # rewrite data/tools.csv from src/tools/npm run tools:csv:check  # exit 1 if data/tools.csv is out of date

Docs

PageWhat it answers
How to add an MCP server to Claude CodeThe claude mcp add command for stdio, HTTP and SSE servers, the three config scopes, and how to debug a server that will not connect.
MCP server for image optimizationHow to let an AI agent compress and convert images (WebP, AVIF, JPEG) through MCP, and the measured savings per format.
MCP tools referenceAll 22 tools, one section each, with the complete input and output schema and an example call.

FAQ

Is Cleanor MCP free?

Yes. It is free to use, with no account, no API key and no paid tier. The hosted endpoint is rate-limited per IP (120 requests per minute in general, 30 per minute for image optimization) so one client cannot starve the rest, and the source is MIT licensed if you would rather run it yourself.

Does it need an API key?

No. Auth is none, which you can verify for yourself at GET https://mcp.cleanor.app/v1/capabilities. There is no signup, no OAuth flow and no token to rotate. Point your client at https://mcp.cleanor.app/mcp and start calling tools.

What data does it send?

Only what you pass to a tool. 21 of the 22 tools are pure functions evaluated on the server, and optimize_image additionally fetches the public image URL you hand it. There is no account, no user identifier, and no database or object store in the server: it is a single Cloudflare Worker with no persistence binding for your inputs. If your data is sensitive, run the stdio build locally with npx -y @cleanor/mcp and nothing leaves your machine.

Can I self-host it?

Yes, in two ways. Run it locally over stdio with npx -y @cleanor/mcp (the same tool registry, with sharp doing the image encoding instead of Cloudflare Images), or deploy this repo to your own Cloudflare account with wrangler deploy, since the Worker entry point and wrangler.jsonc are both in the tree. The Dockerfile builds the stdio server for registries and scanners that prefer to start it themselves.

How is a hosted MCP server different from a local MCP server?

A local (stdio) server is a process your client spawns on your machine: it can reach your filesystem, but you have to install it, keep it updated and carry its dependencies. A hosted (Streamable HTTP) server is just a URL: no install, no version drift, and the heavy work runs on someone else's CPU. Cleanor MCP is offered both ways, and the hosted endpoint is the one to prefer for image optimization, because Cloudflare Images does the encoding and you never install sharp.

Which clients work with it?

Any MCP client that speaks Streamable HTTP: Claude Code, Claude Desktop (as a custom connector), Cursor, VS Code, and the agent frameworks with MCP support. Clients that only speak stdio can run the npm package instead. Both paths expose the identical 22 tools, because both register the same tool array from src/tools/.

Where does the research data come from?

storage_capacity and image_format_savings return numbers from Cleanor Labs' published studies, not from model memory. The format savings come from a controlled benchmark on the 24-image Kodak lossless suite at matched SSIM; the storage numbers come from measured per-item file sizes plus real OS overhead. Every response links its source page on cleanor.app/research.

Related projects

RepoWhat it is
cleanor-app/browser-image-toolsThe in-browser image compression and conversion engine behind cleanor.app.
cleanor-app/image-compressor-chrome-extensionChrome extension: compress, convert and capture images without leaving the page.
cleanor-app/wordpress-image-optimizerWordPress plugin that optimizes and converts your media library.
cleanor-app/cleanor-storage-labThe open benchmarks behind the storage and image-format studies.
cleanor-app/search-indexOpen dataset of monthly search demand, the source for the Cleanor trends studies.

Built by Cleanor Labs

Cleanor is a set of free, private browser tools (files never leave your device) plus original research on device storage and image formats. This MCP server is a thin front door to that toolset and that data, for AI agents.

License

MIT, Cleanor Labs.

Citation

If you use Cleanor MCP or its data, please cite it. Authored by Cleanor Labs, ORCID 0009-0005-4623-961X. The archived, citable version has DOI 10.5281/zenodo.21225551, a concept DOI that always resolves to the latest release. Machine-readable metadata lives in CITATION.cff.

來源:README.md,提交 34f4657

工具

22

optimize_image

Optimize / convert an image for the web

Fetch an image from a public URL and re-encode it smaller (WebP/AVIF/JPEG), optionally resizing to a target width. Returns the optimized image plus before/after byte sizes. Use this when an AI-generated or dropped-in asset (hero image, screenshot, illustration) is too large to ship.

輸入參數widthformatqualityimage_url必填

storage_capacity

How much fits in a phone storage tier

How many photos or minutes of video actually fit in a given storage size, corrected for real OS/filesystem overhead. Backed by Cleanor Labs measured per-item sizes. Use for realistic sample copy, dashboards, or "how many photos fit in 128 GB" answers.

輸入參數contentstorage_gb必填

image_format_savings

Real storage savings of next-gen image formats

How much smaller WebP, AVIF or JPEG XL are than JPEG at matched perceptual quality, from Cleanor Labs’ controlled benchmark. Also reports the "HEIC conversion tax" (converting an iPhone HEIC to JPG/PNG makes it bigger). Use to justify a format choice when building a site or app.

輸入參數formatquality

qr_code

Generate a QR code (SVG)

Encode text or a URL as a QR code and return a crisp, dependency-free SVG you can paste straight into a page, deck or doc.

輸入參數eccsizetext必填

hash

Hash text (SHA family)

Compute a cryptographic hash of text (SHA-1, SHA-256, SHA-384 or SHA-512) and return the hex digest. Use for checksums, cache keys, or verifying content. MD5 is intentionally not offered (broken, and unavailable in Web Crypto). LLMs cannot compute these reliably by hand, so always use this tool instead of guessing.

輸入參數input必填algorithm

uuid

Generate UUIDs

Generate one or more UUIDs. v4 is fully random; v7 is time-sortable (recommended for database keys). LLMs cannot produce cryptographically random or correctly-formatted UUIDs, so always use this tool.

輸入參數countversion

base64

Base64 encode / decode

Encode text to Base64 or decode Base64 back to text (UTF-8 safe). Supports URL-safe alphabet. Use whenever you need to encode/decode data URIs, tokens, or config values instead of guessing the bytes.

輸入參數modeinput必填url_safe

json_format

Format / validate JSON

Validate JSON and pretty-print or minify it, optionally sorting object keys. Returns a precise parse error (with position) if invalid. Use to check and clean JSON instead of eyeballing it.

輸入參數modeinput必填sort_keys

jwt_decode

Decode a JWT (no verification)

Decode a JSON Web Token into its header and payload so you can inspect claims (iss, exp, sub, scopes). The signature is NOT verified and no secret is required or stored. Use to read a token during debugging.

輸入參數token必填

color

Convert a color between formats

Convert a color (hex, rgb() or hsl()) and return hex, RGB and HSL representations at once. Use when picking or translating colors for CSS, design tokens or themes.

輸入參數value必填

slugify

Slugify text for URLs

Turn a title or phrase into a clean, URL-safe slug (lowercase, hyphenated, accents stripped). Use when generating page paths, filenames or anchor IDs.

輸入參數input必填separator

count

Count characters, words, lines, bytes

Accurately count characters (Unicode code points), UTF-16 units, words, lines and UTF-8 bytes in text. LLMs are notoriously bad at counting, so always use this tool for "how many characters/words" questions.

輸入參數input必填

regex_test

Test a regular expression

Test a JavaScript regular expression against sample text and return whether it matches, plus every match with its captured groups and index. Use to verify a pattern instead of reasoning about it in your head. Input and pattern are length-capped to keep it fast and safe.

輸入參數flagsinput必填pattern必填

cron_describe

Explain a cron expression

Parse a standard 5-field cron expression (minute hour day-of-month month day-of-week) into a plain-English breakdown and the next few run times in UTC. Use to sanity-check a schedule instead of guessing what the fields mean.

輸入參數expression必填

unit_convert

Convert between units

Convert a value between units of length, mass, data size, time, speed or temperature. Supported units: mm, cm, m, km, in, ft, yd, mi, nmi, mg, g, kg, t, oz, lb, st, bit, byte, kb, kib, mb, mib, gb, gib, tb, tib, ms, s, min, h, day, week, mps, kph, mph, fps, knot, c, f, k. Use for exact conversions instead of approximating.

輸入參數to必填from必填value必填

datetime

Current or parsed date/time

Get the current date/time, or convert a given timestamp, into a target IANA timezone with ISO, Unix and human-readable forms. Pass a Unix timestamp (seconds or ms) or an ISO string as input; omit it for "now". LLMs cannot know the real current time, so use this instead of guessing.

輸入參數inputtimezone

url_parse

Parse a URL into parts

Break a URL into its components: scheme, host, port, path, decoded query parameters and fragment. Use to inspect or debug a URL instead of parsing it by eye.

輸入參數url必填

base_convert

Convert a number between bases

Convert an integer between number bases 2–36 (e.g. hex to binary, decimal to base-36). Arbitrary precision via BigInt, so large values stay exact. Use for radix conversions instead of doing them by hand.

輸入參數value必填to_basefrom_base

diff

Line diff between two texts

Compute a line-by-line diff between two texts, marking removed lines with "-", added with "+" and unchanged with two spaces, plus a change count. Use to see exactly what changed instead of comparing by eye. Capped at 1000 lines per side.

輸入參數a必填b必填

hmac

HMAC signature of a message

Compute an HMAC (keyed hash) of a message with a secret, using SHA-1/256/384/512, returned as hex or Base64. Use to sign webhook payloads or verify a signature instead of guessing. LLMs cannot compute this by hand.

輸入參數secret必填message必填encodingalgorithm

placeholder_image

Generate a placeholder image (SVG)

Generate a lightweight SVG placeholder image at any size, with an optional label and custom background/text colors. Dependency-free, pastes straight into a page or mockup. Use for wireframes and design stubs instead of hotlinking a placeholder service.

輸入參數bgtextcolorwidthheight

color_palette

Generate a color palette

Build a harmonious color palette from a base color using color-theory rules (complementary, analogous, triadic, tetradic, or monochromatic). Returns each color as hex and HSL. Use to derive a theme or design tokens from one brand color.

輸入參數color必填harmony

版本歷史

1
  1. v0.6.1最新Sep 16, 2026

評分

已滿足適用檢查項的 100%

該評分衡量此收錄條目的完整程度,依據註冊中心條目、端點與公開來源計算得出。它不是對伺服器本身的評價,也不表示伺服器安全或目前可用:可連通性是帶日期的檢查結果,評分也不包含安裝量或使用者評價。

  • 近 30 天內可連通必需25 分
  • 已宣告認證方式必需10 分
  • 已列出工具必需15 分
  • 描述必需15 分
  • 圖示10 分
  • 文件15 分
  • 授權條款10 分