Human Creative AI Workflow & Eval Lab

io.github.jdurey由 GitHub 帳號 jdurey 發布v1.0.0更新於 Nov 14, 2026

Remote MCP tools for 3-vendor AI workflow diagnostics, HAR sandboxes, 50v audits, and rubrics.

發布方自己的服務：TAP

已驗證Streamable HTTP可網頁執行開發者工具安全與監控AI 與機器學習

概覽

AI 產生的概覽

遠端 MCP 伺服器，提供多模型 AI 工作流程診斷、代理評測沙箱、護欄稽核、評審量表校準，以及 SOP 轉 SCORM/QTI 編譯。

可以做什麼
  • 以免費的三模型即時診斷分析 AI 工作流程或職務描述，輸出瓶頸、失效模式、DAG 與上線計畫
  • 依工作流程網址或錄製的 HAR 檔產生離線代理評測沙箱，含任務、獎勵驗證器與基準報告
  • 以 50 組向量進行提示注入與策略繞過壓力測試，回傳加固後的系統提示詞
  • 跨四個模型家族校準評審量表，降低自我偏好偏差與判定翻轉
  • 把 Markdown 或純文字 SOP、操作手冊編譯為 SCORM 1.2 與 QTI 3.0 學習平台套件
適用情境
適合在設計或審查 AI 代理工作流程時取得外部多模型評審、建立評測框架，或對系統提示詞做對抗性檢查。若需把內部 SOP 轉成學習平台可用的課程套件，也可使用。
執行需求
遠端 Streamable HTTP 端點，連線不需本機執行環境、容器或 API 金鑰。診斷工具免費；其餘四個交付工具在 preview_mode 為 true 時可免費預覽，完整交付需透過 Stripe 或 x402 USDC 付費。
安裝前請注意
五個工具中有四個可能產生費用：完整交付分別約 199、149、49 美元，透過 Stripe 或 x402 USDC 支付，除非 preview_mode 為 true，呼叫前請確認模式。提交的工作流程描述、網址、HAR 擷取檔、提示詞與 SOP 文件會傳送到營運方的 Cloudflare 服務及第三方模型供應商。

安裝

在 SourceWeft 中

  1. 開啟 儀表板中的 Human Creative AI Workflow & Eval Lab，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/io.github.jdurey%2Fhuman-creative/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": {
    "human-creative-ai-workflow-eval-lab": {
      "type": "http",
      "url": "https://josh-employer-funnel.josh-durey.workers.dev/mcp"
    }
  }
}

README

Human Creative MCP Server (io.github.jdurey/human-creative)

Zero-install Streamable HTTP Model Context Protocol (MCP) server for multi-vendor AI workflow diagnostics, deterministic HTTP Archive (.har) agent evaluation sandboxes, 50-vector adversarial guardrail audits, 4-family LLM judge rubric calibration, and SOP-to-SCORM/QTI compilation.

  • Remote MCP Endpoint (Streamable HTTP): https://josh-employer-funnel.josh-durey.workers.dev/mcp
  • Registry Namespace: io.github.jdurey/human-creative
  • Architecture Portfolio & Live Diagnostic Hub: https://jdurey.github.io
  • Interactive AI Video Twin (Live Technical Q&A): https://jdurey.github.io/interview/

No local runtime, Docker container, or API key is required to connect and run diagnostics or instant Edge dry-run previews (preview_mode: true).

Available MCP Tools (5)

Tool NamePurposeInstant Free Mode
diagnose_ai_workflowExecutes a live 3-stage cross-family AI workflow or job description diagnostic on Cloudflare Workers AI (Mistral 24B intake -> Llama 3.3 70B 4-node DAG architect + Qwen 2.5 32B independent quality gate). Returns structured bottlenecks, silent failure modes, a 4-node DAG with human checkpoints, a 30-day rollout plan, and matched architecture receipts.Always free (readOnlyHint: true)
build_rea_har_agent_sandboxConverts a live web workflow URL or recorded HTTP Archive (.har v1.2) into an offline deterministic REA AgentEnv MCP sandbox (callable read/write tools + state transitions) paired with 8 multi-step evaluation tasks, 3-vendor Judge-Trust reward verifiers, a 300-DPI Executive Benchmark PDF, and a SHA-256 verified .zip bundle.Set "preview_mode": true
audit_workflow_guardrails_50vRuns a 50-vector multi-family prompt-injection, policy-bypass, and stem-withheld answer-leak stress test judged by 3 independent LLM vendor families, producing a hardened v2.0 system prompt and a 300-DPI Executive Audit PDF.Set "preview_mode": true
calibrate_judge_rubric_4familyEliminates single-model self-preference bias and stochastic verdict flips by decomposing subjective quality criteria into deterministic Python checks paired with a 4-family Judge-Trust rubric panel verified across byte-identical repeats.Set "preview_mode": true
compile_sop_to_scorm_qtiCompiles raw Markdown or plain-text SOPs, runbooks, and training documents into LMS-ready SCORM 1.2 and QTI 3.0 XML mastery packages with stem-withheld blind-QA item verification.Set "preview_mode": true

1-Click Client Setup

Claude Code

Run in your terminal:

bash
claude mcp add --transport http human-creative https://josh-employer-funnel.josh-durey.workers.dev/mcp

Or add to .mcp.json in your project root:

json
{  "mcpServers": {    "human-creative": {      "type": "http",      "url": "https://josh-employer-funnel.josh-durey.workers.dev/mcp"    }  }}

Cursor

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

json
{  "mcpServers": {    "human-creative": {      "url": "https://josh-employer-funnel.josh-durey.workers.dev/mcp"    }  }}

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

json
{  "mcpServers": {    "human-creative": {      "serverUrl": "https://josh-employer-funnel.josh-durey.workers.dev/mcp"    }  }}

Cline

Add to cline_mcp_settings.json:

json
{  "mcpServers": {    "human-creative": {      "type": "streamableHttp",      "url": "https://josh-employer-funnel.josh-durey.workers.dev/mcp",      "disabled": false    }  }}

Quick Verification (curl)

List all live tools via JSON-RPC 2.0:

bash
curl -s -X POST https://josh-employer-funnel.josh-durey.workers.dev/mcp \  -H "Content-Type: application/json" \  -d '{"jsonrpc":"2.0","id":1,"method":"tools/list","params":{}}'

Run a free 3-vendor workflow diagnostic:

bash
curl -s -X POST https://josh-employer-funnel.josh-durey.workers.dev/mcp \  -H "Content-Type: application/json" \  -d '{    "jsonrpc": "2.0",    "id": 2,    "method": "tools/call",    "params": {      "name": "diagnose_ai_workflow",      "arguments": {        "mode": "workflow",        "org_hint": "Example Engineering Team",        "input": "Our customer support triage agent hallucinates refund policy exceptions and lacks deterministic pre-send verification."      }    }  }'

Links & Contact

License

MIT License. See LICENSE.

來源：README.md，提交 056a68f

工具

5

diagnose_ai_workflow

Live 3-Vendor AI Workflow & JD Diagnostic (Free)

FREE live 3-stage cross-family AI workflow and JD diagnostic executed immediately on Cloudflare Workers AI (Mistral 24B intake -> Llama 3.3 70B 4-node DAG architect + Qwen 2.5 32B independent quality gate). Returns structured bottlenecks, silent failure modes, a 4-node DAG with human checkpoints, a 30-day rollout plan, and matched portfolio receipts.

輸入參數modeinput必填org_hint

build_rea_har_agent_sandbox

Build Offline REA HAR-to-AgentEnv Sandbox ($299 Stripe / $3.50 x402 USDC / Free Preview)

Converts a live web workflow URL or recorded HTTP Archive (.har v1.2) into an offline deterministic REA AgentEnv MCP sandbox (callable read/write tools + state transitions) paired with 8 multi-step evaluation tasks, 3-vendor Judge-Trust reward verifiers, a 300-DPI Executive Benchmark PDF, and a SHA-256 verified .zip bundle ($299 Stripe / $3.50 x402 USDC, or set preview_mode=true for an instant free Edge dry-run).

輸入參數target_urlclient_name必填preview_modecontact_emailpayment_tokenworkflow_name必填observed_endpoints

audit_workflow_guardrails_50v

50-Vector Multi-Family Guardrail & Leak Stress-Test ($199 / Free Preview)

Runs a 50-vector multi-family prompt-injection, policy-bypass, and stem-withheld answer-leak stress test judged by 3 independent LLM vendor families, producing a hardened v2.0 system prompt and 300-DPI Executive Audit PDF ($199 Stripe / $2.50 x402 USDC, or set preview_mode=true for a free 5-vector Edge dry-run).

輸入參數client_name必填preview_modecontact_emailpayment_tokenbot_or_workflow_name必填system_prompt_or_rules必填

calibrate_judge_rubric_4family

4-Family Judge-Trust Rubric Calibration ($149 / Free Preview)

Eliminates single-model self-preference bias and stochastic verdict flips by decomposing subjective quality criteria into deterministic Python checks + a 4-family Judge-Trust rubric panel verified across byte-identical repeats ($149 Stripe / $2.00 x402 USDC, or set preview_mode=true for a free calibrated rubric preview).

輸入參數client_name必填task_domain必填preview_modecontact_emailpayment_tokencurrent_rubric必填

compile_sop_to_scorm_qti

Compile SOP to Verified SCORM 1.2 & QTI 3.0 Packages ($49 / Free Preview)

Compiles raw Markdown/text SOPs, runbooks, or training docs into LMS-ready SCORM 1.2 + 1EdTech QTI 3.0 XML mastery packages with stem-withheld blind-QA item verification ($49 Stripe / $0.75 x402 USDC, or set preview_mode=true for a free Knowledge Component + QTI 3.0 XML item preview).

輸入參數client_namecourse_title必填preview_modesop_markdown必填contact_emailpayment_token

版本歷史

1
  1. v1.0.0最新Nov 14, 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

其他 Human 整合

TAP

tech.human官方發布者

TAP 是託管式遠端 MCP 端點，透過佔位密鑰、政策檢查與可選的人工核准，為 AI 代理隔離憑證。

Streamable HTTP可網頁執行驗證方式
開發者工具需要憑證Oct 24, 2026

UGC, the Human Ad Network

africa.ugc官方發布者

UGC（「Human Ad Network」）的遠端 MCP 端點，讓助理在對話中處理廣告活動、剪輯與 Clip Studio。

Streamable HTTP可網頁執行驗證方式
商業與電商需要憑證Oct 24, 2026

A Human For Hire

com.ahumanforhire

讓助理雇用真人：瀏覽付費任務目錄，透過 x402 在 Solana 上支付 USDC，發布任務並輪詢回覆。

STDIO僅桌面
金融6 個工具Nov 14, 2026

HumanCalls

由 GitHub 帳號 Hire-Overseas 發布

讓助理委託真人撥打美國或加拿大商家電話，並取回通話摘要、答覆、費用、錄音與逐字稿。

Streamable HTTP可網頁執行驗證方式
效率與工作流需要憑證Nov 12, 2026

Become Human

由 GitHub 帳號 DanielDDHM 發布

本機 MCP 伺服器，為助理提供關於你工作方式的長期画像，以及依專案記錄的想法地圖，並以可編輯的 Markdown 檔案保存。

STDIO僅桌面
效率與工作流8 個工具Nov 12, 2026