
OpenClaw FinOps
io.github.maryadawson-code由 GitHub 帳號 maryadawson-code 發布v1.0.0更新於 Oct 25, 2026
Verified cloud cost forecasting for AI agents. AWS, GCP, Azure pricing matrix.
發布方自己的服務:OpenClaw Direct
概覽
一個遠端 MCP 伺服器,為助理提供經核實的 AWS、GCP 與 Azure 價格,用來預測雲端部署成本而非憑空猜測。
- 可以做什麼
- 針對 AWS、GCP 或 Azure 上的一組雲端服務預測每月部署成本,提供逐項明細與總估算(forecast_deployment_cost)。
- 適用情境
- 當助理在規劃或比較雲端部署、需要對支援的執行個體、資料庫、快取與儲存服務取得有依據的每月成本估算,而非模型猜測時使用。
- 執行需求
- 遠端 streamable HTTP 端點,不需本機執行環境或安裝套件。需要 API 金鑰,透過 x-api-key 標頭傳送。免費額度為每月 25 次操作,超出需付費升級。
安裝
在 SourceWeft 中
- 開啟 儀表板中的 OpenClaw FinOps,將其新增到工作區。
- 連接 x-api-key 憑證。它們會被私密儲存,絕不會在公開頁面上展示。
- 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.maryadawson-code%2Fopenclaw-finops/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。其他 MCP 客戶端
把它新增到你客戶端的 mcpServers 設定中。
{
"mcpServers": {
"openclaw-finops": {
"type": "http",
"url": "https://openclaw-finops.marywomack.workers.dev/mcp",
"headers": {
"x-api-key": "<YOUR_API_KEY>"
}
}
}
}README
IntegrityPulse FinOps
Stop your AI agents from hallucinating cloud costs. Get real pricing forecasts inside the conversation.
[MCP Server] [Cloudflare Workers] [License: MIT]
What is this?
IntegrityPulse FinOps is a remote MCP server that gives AI coding agents accurate, real-time cloud deployment cost forecasts. Instead of your agent guessing that "an EC2 instance costs around $50/month," it calls a tool backed by a verified pricing matrix and returns a line-item breakdown.
One tool. Three providers. Zero hallucinations.
Quick Start
Claude Desktop
Add this to your claude_desktop_config.json:
Cursor
Add to your .cursor/mcp.json:
cURL
The Tool: forecast_deployment_cost
Supported services:
Why It Exists: FinOps for the Agentic Era
AI agents are moving from "write me code" to "deploy this for me." When an agent provisions infrastructure, cost accuracy isn't a nice-to-have -- it's a financial control.
The problem: LLMs hallucinate pricing. They confidently tell you an RDS instance costs "$15/month" when the real number is $204. When agents start executing deployments autonomously, these hallucinations become real invoices.
The solution: IntegrityPulse FinOps is a grounded pricing oracle that agents call as a tool. The pricing matrix is maintained, versioned, and deterministic. No generation, no guessing.
Agentic Commerce & the Revenue Gate
This server is also a reference implementation of Revenue-Gated MCP -- a pattern for monetizing MCP tools without breaking the agent experience.
Here's how it works:
- Every request carries an
x-api-keyheader. - The server checks the user's
tierandmonthly_usage_countin Supabase. - Free tier users get 25 operations/month.
- When the limit is hit, the server does NOT return an HTTP error. Instead, it returns a valid MCP tool result with
isError: true:
Why isError: true instead of HTTP 402/429?
Because the consumer is an LLM, not a browser. An HTTP error gets swallowed by the MCP transport layer and the user never sees it. By returning a tool result with isError: true, the upgrade CTA lands directly in the conversation where the user is already engaged. The agent reads it, surfaces it, and the user can act on it immediately. This is commerce at the point of intent.
Stripe Integration
When a user upgrades via the billing link, Stripe fires a checkout.session.completed webhook to /api/webhook/stripe. The server automatically:
- Updates their tier from
FREEtoPRO - Resets their usage counter to 0
- The next agent request goes through seamlessly
Architecture
- Runtime: Cloudflare Workers (edge, ~0ms cold start)
- Framework: Hono (lightweight, Workers-native)
- MCP SDK:
@modelcontextprotocol/sdkwith Streamable HTTP transport - Auth: API key in
x-api-keyorAuthorization: Bearerheader - Database: Supabase (PostgreSQL)
- Payments: Stripe Checkout + Webhooks
Self-Hosting
License
MIT
來源:README.md,提交 72abbde
工具
0版本歷史
1- v1.0.0最新Sep 16, 2026
評分
已滿足適用檢查項的 75%
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15 分
請讓伺服器在無憑證時回應 tools/list,或發布可在沙箱中無憑證啟動的軟體包。
- 描述必需15 分
- 圖示10 分
- 文件15 分
- 授權條款10 分
請在儲存庫或註冊中心條目中宣告授權條款。
其他 Openclaw 整合
OpenClaw Direct
direct.openclaw官方發布者
一個遠端 MCP 端點,讓助理用自然語言部署、監控與管理 OpenClaw AI 助理。

OpenClaw MCP Ecosystem
由 GitHub 帳號 yedanyagamiai-cmd 發布
一個部署在 Cloudflare Workers 上的遠端端點,宣稱捆綁九個 MCP 伺服器,涵蓋安全掃描、知識圖譜、合規、金融、記憶、分析等。

Openclaw Upgrade Orchestrator Mcp
由 GitHub 帳號 temurkhan13 發布
一個本機 MCP 伺服器,用於 AI 執行環境升級安全,透過快照比對與指紋偵測供應商端的回歸問題。

Openclaw Skill Vetter Mcp
由 GitHub 帳號 temurkhan13 發布
一個本機 MCP 伺服器,讓助理審查第三方 AI 代理擴充套件,以及 AGENTS.md、.cursor/rules、CLAUDE.md 等代理設定檔。

Openclaw Output Vetter Mcp
由 GitHub 帳號 temurkhan13 發布
本機 MCP 伺服器,用來核對 AI 代理的說法是否屬實,提供內嵌依據檢查、例外被吞掉偵測與對話紀錄審查。