Spec Driven Development
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
作者提供的说明(英文原文)
Check if your MCP server is ready for MCP directories, then get it submitted.
这三项永远不会合并成一个分数:意见、测量和人气是不同的东西。
我们还没有为它写通俗的介绍。下面是我们从注册中心读取到的事实。
选择你使用的 AI 应用,然后复制设置。
你需要
你用的是哪个应用?
在上方选择你的应用。我们会记住它,下次按钮会显示“安装到 …”。
适用于网页版、桌面版和手机版 Claude:
https://mcpserverlist.io/mcp在终端中运行这条命令
claude mcp add --transport http mcpserverlist https://mcpserverlist.io/mcp粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"mcpserverlist": {
"url": "https://mcpserverlist.io/mcp"
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"mcpserverlist": {
"type": "http",
"url": "https://mcpserverlist.io/mcp"
}
}
}浏览器会先询问是否打开应用,然后应用会显示设置并请你确认。密钥的值会留空,由你自己填写。
ChatGPT 只能连接互联网上的服务器:
https://mcpserverlist.io/mcp菜单名称可能因套餐和版本不同而略有差异。
你的选择通过 Cookie 保存在这台设备上,无需账户。
这些设置是根据作者在注册中心的条目生成的。如果 README 的写法不同,请以 README 为准。
工具名称和描述与服务器在 2026年11月6日 报告的完全一致。这些内容由作者撰写,不是我们写的;我们以纯文本形式显示。
list_directoriesAnonymous catalog with source references and requirements. Discovery does not establish purchase eligibility; run a full authenticated preflight.
check_repositoryAnonymous free diagnostic for a public GitHub repository. This is not the per-destination purchase preflight.
create_accountAnonymous signup returns a new account API key once. Save it securely; email is not an access or recovery credential.
get_accountRequires API key. Read account and configured billing/execution availability. Generic launch credits are not sold.
connection_instructionsRead-only instructions for direct HTTPS credential upload. Never put third-party secrets in chat or MCP tool arguments.
list_connectionsRequires API key. Returns owned connection IDs, providers, expiry and revocation metadata. Never returns plaintext or ciphertext.
revoke_connectionRequires explicit owner authorization. Revoking a connection prevents jobs using it from completing until reconnected.
list_serversRequires API key; returns only your saved profiles.
create_serverRequires API key. Save accurate public profile fields before checking all selected destination prerequisites. Smithery currently needs a public unauthenticated Streamable HTTP endpoint and owned namespace/server name.
import_repositoryInspect a public repository and save an owned launch intake. Infers evidenced public facts and asks only missing facts. No upstream writes or payment.
get_launch_intakeRead your persisted profile, evidence, missing questions and next action.
answer_launch_questionsSave explicit public answers to the questions in an owned intake. No upstream mutation or payment.
check_launch_coverageCheck all implemented destinations with read-only provider requests, remove blocked destinations and preserve prerequisites. Destinations whose topic or eligibility does not fit this server come back as state not_applicable (not a failure; reasons say how to set directory_categor
prepare_repositoryFree owner-authenticated inspection at a pinned GitHub revision. Returns exact proposed file diffs, preservation warnings and approval digest. No external writes.
get_repository_preparationRead owned pinned patch, warnings and digest before authorizing a PR within existing owner authority or requesting missing approval.
create_preparation_pull_requestRequires explicit owner approval of the exact displayed patch digest. Creates a branch and PR in the source repository, never merges or publishes. After uncertain writes, only reconciles matching existing PRs; never blindly retries writes.
create_publisher_taskFree preparation for Claude/OpenAI owner review. Saves a field packet and eligible package; never submits, proves delivery or adds paid coverage. No secrets or credential uploads.
list_publisher_tasksRead owned preparation tasks and operator journal status; pending provider work is not fulfillment.
get_publisher_taskRead the exact owned task revision, digest, missing owner/provider actions and optional authenticated ZIP download link. No raw binary or credentials are returned through MCP.
authorize_publisher_taskOnly use within existing explicit owner authority covering this exact task revision, digest, submit/publish action and truthful attestations; obtain missing authority first. Records a guarded grant, not proof of submission, approval or publication. Does not pay.
list_form_tasksRead free prefilled packets (submit URL, exact field values, which free option to choose) for directories that only accept a web form. The owner submits each form; nothing is sent, paid or verified by this service.
mark_form_task_submittedRecords only that the owner says they submitted the directory form themselves. Not proof of acceptance or listing.
run_preflightRequires API key. Read-only provider checks produce an immutable 30-minute report, exact public payloads, blockers and remedies. No external submission or charge occurs. Unsupported/paused destinations cannot be bought.
get_preflightRequires API key. Review every target, blocker and proposed public payload with the owner before authorizing a paid launch.
create_launchRequires API key and explicit owner authorization of the displayed preflight plans. Creates an unpaid draft bound to a ready owned preflight; this call does not pay. Verified payment later automatically queues the authorized submissions.
payment_instructionsRead-only protocol instructions for an approved launch_id. No payment is made by this tool. Obtain owner authorization for the saved launch amount (current new-launch offer $49); payment starts automatic submissions.
create_checkoutRequires API key, launch ID and stable purchase key. Human payment fallback for the saved immutable service fee (current new-launch offer $49); only verified paid settlement queues jobs. This tool does not autonomously pay.
list_service_casesRead owned delivery reviews, assigned operator, due date and fulfillment summary.
get_service_caseRead an owned review with operator updates and resolution evidence.
reply_to_service_caseAdd a customer reply or acknowledgement. Never include credentials or private payloads.
replace_launch_connectionAfter the owner securely uploads a fresh connection, verify repository ownership and replace the same already-authorized provider for a paid launch. Cannot retry uncertain writes or change submission plans.
get_orderRequires API key. Returns your owned order and bound launch after payment or timeout. Do not purchase again while outcome is unknown.
list_launchesRequires API key. Read durable execution and provider result states.
get_launchRequires API key. Read owned jobs, real provider IDs, evidence, ingestion waits and operator-attention states. Submitted is not approved; listed requires matching provider evidence.
get_launch_overviewRequires API key. One owner-scoped summary of a launch: totals, the refund-commitment promise (qualifying, listed, needed, on track), what is waiting on the owner, and every destination with state, typical review time, links and timeline. Listed always means verified; submitted i
我们检查的是任何人都能看到的信息;我们没有阅读代码。不适用的检查项不计入总数。 · 检查于 2026年11月7日
注册名称已通过 GitHub 账户 sudomichael 验证,该账户拥有此仓库
我们还没有从 GitHub 读取这个仓库
我们还没有从 GitHub 读取这个仓库
在 225 毫秒内响应(HTTP 200)
无需登录
已读取 35 个工具描述,未发现针对 AI 的隐藏指令
与注册中心里这个分类中所有活跃的服务器比较。只有当发布者拥有该仓库时,星标数才计入。
它在 3 项适用的基础检查中通过了 3 项。我们没有阅读代码。在添加密钥或令牌之前,请先看看这个服务器要求哪些权限,并使用只具备所需权限的密钥。
注册中心没有列出必需的密钥,不过有些服务器会在首次使用时要求你登录。
它在作者的云端运行。通过互联网连接,无需安装,但你发送的内容会经过作者的服务器。
GitHub 仓库最后更新于–,最新版本(0.1.0)于3天前进入注册中心。
支持 MCP 的应用,例如 Claude Desktop、Claude Code、Cursor 和 VS Code。它通过 URL 连接,所以接受远程 MCP 服务器的应用也可以使用,例如开启开发者模式的 ChatGPT。
同一分类中有:Spec Driven Development, Remote Desktop Commander, Qemu MCP
同一分类中其他作者的服务器,按星标数排序。
如何看卡片:★ GitHub 星标i三大支柱:评价 · 检查 · 使用情况i更新 = 最新版本发布到注册中心的时间i
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.
Boot and drive QEMU VMs from an AI agent: keys, screenshots, serial. No guest agent.
按原文从作者的仓库复制。PopNx MCP 没有审查或背书其内容。HTML、图片和脚本已被移除。链接会离开本站。
Stdio MCP server for MCP Server List. It connects to the hosted service at https://mcpserverlist.io/mcp and passes tools through unchanged, so any stdio-only client or directory can use it.
It helps MCP server authors get their server listed in MCP directories. It checks whether your server is ready to submit, and offers a $99 managed submission service that handles the submissions for you.
Run with npx -y mcpserverlist-mcp (Node 20 or newer).
Claude Desktop (claude_desktop_config.json), Cursor (~/.cursor/mcp.json):
{
"mcpServers": { "mcpserverlist": { "command": "npx", "args": ["-y", "mcpserverlist-mcp"] } } }
Claude Code:
claude mcp add mcpserverlist -- npx -y mcpserverlist-mcp
To use an API key, add "env": { "MCP_SERVER_LIST_API_KEY": "..." } to the config above, or pass -e MCP_SERVER_LIST_API_KEY=... to claude mcp add.
Clients that support Streamable HTTP can connect straight to the hosted server:
claude mcp add --transport http mcpserverlist https://mcpserverlist.io/mcp
Or in a JSON config, { "mcpServers": { "mcpserverlist": { "url": "https://mcpserverlist.io/mcp" } } }.
| Variable | Required | Description |
|---|---|---|
MCP_SERVER_LIST_API_KEY | No | Sent as Authorization: Bearer . Only sent when set. |
MCP_SERVER_LIST_URL | No | Remote endpoint. Default https://mcpserverlist.io/mcp. |
Discovery and check tools work without a key. Tools that act on your account (your servers, submissions) need an API key, which you get by calling the create_account tool. Use tools/list for the current set.
tools/list and tools/call unchanged. Resources and prompts are forwarded only if the remote advertises them. The remote's server instructions are mirrored.npm install
npm test # builds, then runs tests against an in-process fake remote
MIT