MCP Compress
List compression profiles, estimate gains, and compress images, PDFs, and Office files.
我該用這個嗎
品質與安全性
根據工具定義與協定合規性的自動化分析。
上下文成本
這是每次將伺服器的工具載入模型上下文時所消耗的約略 token 數量。數量越高，可用於其他工作的注意力就越少。
安裝
一鍵安裝
將以下內容加入你的 `claude_desktop_config.json` 檔案：
{
"mcpServers": {
"mcp-compress": {
"url": "https://mcp-compress.custom-digital-services.com/mcp"
}
}
}遠端端點
https://mcp-compress.custom-digital-services.com/mcpstreamable-http它能做什麼
工具清單
工具（3）
🟢list_compression_profiles
List the three compression presets (light, balanced, maximum) with their exact image quality and max-dimension caps, a note per preset, and the maximum accepted input size. Values are kept in parity with the Compresseur Local web app. Takes no parameters.
輸入結構描述
{
"type": "object",
"properties": {},
"$schema": "http://json-schema.org/draft-07/schema#"
}🟡estimate_compression(fileBase64, fileName, level)
Detect a file (family, format, size from magic bytes) and estimate the compression gain by ACTUALLY compressing it. Pass the file as base64 (fileBase64) plus its fileName (the extension disambiguates Office formats). Omit 'level' to estimate all three presets, or set it for a single one. Returns the detected metadata and, per level, the resulting byte size, gain percentage and output format. Max input size: 26214400 decoded bytes; the HTTP body must accept base64 (decoded x 4/3 plus margin).
輸入結構描述
{
"type": "object",
"properties": {
"fileBase64": {
"type": "string",
"minLength": 1
},
"fileName": {
"type": "string",
"minLength": 1
},
"level": {
"type": "string",
"enum": [
"light",
"balanced",
"maximum"
]
}
},
"required": [
"fileBase64",
"fileName"
],
"additionalProperties": false,
"$schema": "http://json-schema.org/draft-07/schema#"
}⚪compress_file(fileBase64, fileName, level, custom)
Compress a file and return the result as base64. Pass fileBase64 + fileName. Choose EITHER a preset 'level' (light|balanced|maximum, default balanced) OR a 'custom' profile { imageQuality: 1-100, maxDimension?: px, convertFormat?: webp|avif|jpeg } -- supplying both is rejected. Supported families: images (PNG/JPG/WebP/AVIF), PDF (embedded JPEG images re-encoded, no AGPL tool), and Office OOXML (DOCX/PPTX/XLSX embedded media re-encoded). Returns fileBase64, bytesBefore, bytesAfter, gainPercent, mimeType and the output format. If compression would enlarge a non-converted file, the original is returned with 0% gain. WARNING: convertFormat 'avif' is CPU/RAM heavy (>= 1 GB RAM for large images). Max input size: 26214400 decoded bytes.
輸入結構描述
{
"type": "object",
"properties": {
"fileBase64": {
"type": "string",
"minLength": 1
},
"fileName": {
"type": "string",
"minLength": 1
},
"level": {
"type": "string",
"enum": [
"light",
"balanced",
"maximum"
]
},
"custom": {
"type": "object",
"properties": {
"imageQuality": {
"type": "integer",
"minimum": 1,
"maximum": 100
},
"maxDimension": {
"type": "integer",
"minimum": 16,
"maximum": 20000
},
"convertFormat": {
"type": "string",
"enum": [
"webp",
"avif",
"jpeg"
]
}
},
"required": [
"imageQuality"
],
"additionalProperties": false
}
},
"required": [
"fileBase64",
"fileName"
],
"additionalProperties": false,
"$schema": "http://json-schema.org/draft-07/schema#"
}社群
證據