Farpy

io.github.Mangomunchrv1.1.0更新於 Oct 10, 2026

FARPY routes GPU workloads to qualified compute and returns verified artifacts and receipts.

已驗證Streamable HTTP可網頁執行Cloud & InfrastructureMedia & Design

概覽

AI 產生的概覽

提交有界的 GPU 工作負載(例如 Blender Cycles 算圖),回傳可下載的產出物與可驗證的執行憑證。

功能
FARPY 接受有界的 GPU 工作負載,以非同步方式執行,並交付完成的產出物。其公開 MCP 工具為 inspect、quote、start、status、download 與 receipt。目前唯一公開的工作負載是使用 .blend 輸入的 Blender Cycles GPU 算圖,依完成的影格計費,並提供 SHA-256 產出物驗證與執行憑證作為佐證。
適用情境
當助理需要在不自備 GPU 硬體的情況下,把 GPU 算圖或類似的有界運算工作外包出去,且需要為完成的產出物取得可驗證憑證時,適合使用。此服務處於公開 alpha 階段,適合評估而非關鍵正式用途。
執行需求
遠端 MCP 端點 Node.js 從 mcp 目錄建置的本地 stdio 伺服器。本地伺服器需要環境變數 FARPY_AGENT_TOKEN(farpy_agent_ 開頭的 Job API 金鑰),所有工具皆必需,缺少時會直接失敗。FARPY_API_BASE 為選用,可覆寫 API 來源。需要網路存取。
安裝前請注意
工作會產生費用:公開工作負載依完成的影格計費,start 工具是消費界線,需要 legal_acceptance=FARPY_LEGAL_V1。FARPY_AGENT_TOKEN 憑證為所有工具所必需,應妥善保密。提交工作負載會把輸入檔案傳送給第三方服務,start、download 與 receipt 操作的是遠端工作而非本地資料。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 Farpy,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。

其他 MCP 客戶端

把它新增到你客戶端的 mcpServers 設定中。

{
  "mcpServers": {
    "farpy": {
      "type": "http",
      "url": "https://api.farpy.com/mcp"
    }
  }
}

README

FARPY

Verified GPU compute for humans, applications, and agents.

FARPY accepts bounded GPU workloads, executes them asynchronously, delivers completed artifacts, and issues verifiable execution receipts.

Public workload

Blender Cycles GPU rendering — $0.01 per completed frame.

Current public contract:

  • Workload: Blender Cycles GPU
  • Input: .blend
  • Execution: asynchronous
  • Output: downloadable artifact
  • Integrity: SHA-256 artifact verification
  • Evidence: FARPY execution receipt
  • Transparency: public proof surfaces
  • Status: LIVE_PUBLIC_ALPHA

Additional workloads are under development and are not currently public.

Agent / API access

Typical lifecycle:

discover → price → submit → JOB-ID → execute → download → verify receipt

Agent discovery

Public surfaces

Local MCP / Glama

A local stdio MCP server for Cursor, Claude Desktop, Glama, and other MCP clients lives in mcp/.

It is a thin public client over FARPY's existing Job API. It does not include production server code and never logs FARPY_AGENT_TOKEN.

Build and run

bash
cd mcpnpm installnpm run buildnpm testnode dist/mcp-server.js

Environment (see .env.example):

  • FARPY_AGENT_TOKEN — Job API farpy_agent_ key. Required for every tool. The server fails closed if it is missing.
  • FARPY_API_BASE — API origin. Default: https://api.farpy.com

Example local client config:

json
{  "mcpServers": {    "farpy": {      "command": "node",      "args": ["/absolute/path/to/farpy-public/mcp/dist/mcp-server.js"],      "env": {        "FARPY_AGENT_TOKEN": "farpy_agent_..."      }    }  }}

Public MCP V1 tools: inspect, quote, start, status, download, receipt. inspect and quote do not spend. start is the spend boundary and requires legal_acceptance=FARPY_LEGAL_V1.

Scope

FARPY is currently a public-alpha service.

Public capabilities, pricing, and limits are authoritative. Historical research, internal tests, and experimental workloads do not imply current public availability.

來源:README.md,提交 223f9da

工具

0
工具後設資料尚未被收錄。

版本歷史

1
  1. v1.1.0最新Sep 16, 2026