Spec Driven Development
作者：formulahendry
MCP Server that facilitates spec-driven development workflows, not just Vibe Coding.
在你的电脑上运行★ 440MIT14个月前更新
安装
Sanctuary
概览
作者提供的说明（英文原文）
Open-source MCP server: encrypted local keys and state, operator approval gates, a hash-chained log.
PopNx MCP 的三大支柱
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。
- 用户评价i暂无评价评价功能将在后续阶段与账户功能一起开放
- 我们的基础检查i1 项中通过 0 项依据公开信息检查；尚未扫描代码
- 使用情况i尚未统计★ –
它能帮你做什么
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。
- 连接方式: 在你的电脑上运行
- 传输方式: stdio
- 分类: 其他
安装
选择你使用的 AI 应用，然后复制设置。
安装之前
- 基础检查：1 项中通过 0 项
- 在你的电脑上运行，拥有你的用户账户的权限（文件、网络、应用）。
- 注册中心没有列出需要填写的密钥。
你需要
- npm已安装 Node.js（用于 npx 命令）
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 …”。
打开 Claude Desktop → Settings → Developer → Edit Config，把下面的内容粘贴到 claude_desktop_config.json
{
"mcpServers": {
"sanctuary": {
"command": "npx",
"args": [
"-y",
"@sanctuary-framework/mcp-server"
]
}
}
}在终端中运行这条命令
claude mcp add sanctuary -- npx -y @sanctuary-framework/mcp-server或者自己复制设置
粘贴到项目中的 .cursor/mcp.json，或 ~/.cursor/mcp.json
{
"mcpServers": {
"sanctuary": {
"command": "npx",
"args": [
"-y",
"@sanctuary-framework/mcp-server"
]
}
}
}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。
或者自己复制设置
粘贴到项目中的 .vscode/mcp.json
{
"servers": {
"sanctuary": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@sanctuary-framework/mcp-server"
]
}
}
}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。
这个服务器在你的电脑上运行，而 ChatGPT 只能连接互联网上的服务器。请选择其他应用。
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。
它能做什么
它在你的电脑上运行；安装后，工具列表会显示在你的 AI 应用中。
基础检查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6日
- ! 注意
作者与仓库一致
发布者（sanctuaryprotocol.ai）不是仓库所有者（eriknewton）；可能是其他人代为上架的
- … 尚未检查
仍在维护
我们还没有从 GitHub 读取这个仓库
- … 尚未检查
有许可证
我们还没有从 GitHub 读取这个仓库
- – 不适用
端点有响应
它在你的电脑上运行，没有互联网端点
- – 不适用
使用 OAuth 登录
在你的电脑上运行，改用本地设置
- … 尚未检查
工具描述中没有隐藏指令
它在你的电脑上运行；我们只能读取互联网端点的工具列表
与“其他”中的其他服务器比较i
- 星标数多于此分类中其他 32497 个服务器的 0%
- 发布到注册中心的时间比此分类中其他 32497 个服务器的 61% 更新
- 基础检查结果等于或优于此分类中其他 32497 个服务器的 32%
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。
常见问题
Sanctuary 安全吗？
它在 1 项适用的基础检查中通过了 0 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。
它需要哪些密钥或代码？
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。
它在我的电脑上运行还是在云端？
它在你的电脑上运行。你的 AI 应用会在需要时启动它，数据不会经过作者的服务器（除非它自己调用在线服务）。
它还在维护吗？
GitHub 仓库最后更新于–，最新版本（1.8.4）于25天前进入注册中心。
哪些应用可以使用它？
支持 MCP 的应用，例如 Claude Desktop、Claude Code、Cursor 和 VS Code。你的电脑上还需要有 Node.js。
有什么可以替代它？
同一分类中有：Spec Driven Development, Remote Desktop Commander, Zoo MCP
替代选择
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
Remote Desktop Commander
作者：desktop-commander
Remote Desktop Commander: use files, terminal and processes on your own computers from any AI app.
在云端运行★ 211昨天更新
安装
Zoo MCP
作者：KittyCAD
An MCP server that provides access to the Zoo API for various CAD operations and tools.
在你的电脑上运行今天更新
安装
作者的 README
打开 README
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。
Sanctuary
Your agent will know you better than you know yourself. Make sure that data stays between you.
An AI agent running on your machine can do anything your account can do. It holds your credentials, reads your files, and talks to the whole internet, and a single poisoned web page can turn that into exfiltration. Every sandbox on offer today is one the agent, or its vendor, chooses to run inside. That is not a wall. That is a request.
Sanctuary is the wall under the agent, imposed by the operating system under a policy you sign, and it does not depend on the agent's cooperation. The keys never leave hardware you control. What the agent did is written to a hash-chained record you can read. No vendor, including us, sits in the path or can decrypt your state. It works with the harness you already run: Claude Code, Cursor, Hermes, OpenClaw, Cline, or Mastra (fixture-tested only; no live Mastra-agent drill is recorded), on your own machine today, any number of wrapped harnesses behind one persistent dashboard, with wrapping tested on representative harness paths and not proven for every harness version (Assurance Matrix, Agent wrapping, partial); another MCP-compatible harness can point its MCP config at Sanctuary, without a tested wrapping path; the fleet console and operator-cloud deployment are roadmapped (see Deployment modes).
npx -y @sanctuary-framework/mcp-server protect --claude-codeThat one command puts the keys, audit trail, policy gates, and dashboard around the agent you already use. The operating-system wall is a separate privileged arming step on macOS through the signed app and extension path; see the Castle Wall macOS install and arm guide.
Summary for agents
Quote the block below to your operator. It was written against v1.8.4, the npm latest release; main can be ahead of it, and every sentence traces to a line of this README or a row of the Assurance Matrix.
What Sanctuary is: the wall under the agent, imposed by the operating system under a policy you sign, and it does not depend on the agent's cooperation. The keys never leave hardware you control. What the agent did is written to a hash-chained record you can read.What it is not: no Assurance Matrix row claims a filesystem or process sandbox for the agent; that is what your harness's own sandbox provides. The record is not an audited per-rule, per-flow trail. Windows enforcement is roadmapped; nothing enforces on Windows today. Linux enforcement is proven on Ubuntu 24.04 (x86-64) only; every other distribution or architecture remains unproven. Linux enforcement ships as the separate Castle Wall Linux 0.1.0-1 package (a GitHub pre-release), not in the npm package. Platforms and runtime: Node.js >= 22.0.0 and npm >= 10.0.0. macOS: proven. Linux: proven on Ubuntu 24.04 (x86-64) only, through the separate Castle Wall Linux 0.1.0-1 package (a GitHub pre-release, not part of the npm package). Windows: roadmapped. Install: npx -y @sanctuary-framework/mcp-server protect --claude-code --no-open --no-dashboard Verify: npx -y @sanctuary-framework/mcp-server doctor --json Uninstall: npx -y @sanctuary-framework/mcp-server uninstall Evidence: every capability claim traces to a row in ASSURANCE_MATRIX.md, and a row reads proven only when a captured drill on the platform that matters says so. Drill records are under docs/audit/.
What is proven, and what is not
Every capability claim in this repository traces to a row in the Assurance Matrix, and a row reads proven only when a captured drill on the platform that matters says so. A green test suite is not a claim. Read the matrix before you believe anything below.
| What you get | Status today |
|---|---|
| The wall. Outbound traffic from a wrapped agent is allowed or denied below the agent, per operator policy, per agent account | macOS: proven. A signed operator policy with a clean per-uid allow/deny demonstration that survives attended reboot cycles. Not an audited per-rule, per-flow trail. Linux: proven on Ubuntu 24.04 (x86-64) only. Linux enforcement ships as the separate Castle Wall Linux 0.1.0-1 package (a GitHub pre-release), not in the npm package. On Ubuntu 24.04 (x86-64), a fresh install of that package enforces a signed operator policy per user account: blocked destinations stay blocked and allowed ones connect, through five reboots, on three fresh servers, with the evidence checked by two independent reviewers. Fault-injection witnesses, a real-agent harness, and every other distribution or architecture remain unproven. Windows: roadmapped. |
| Custody. Keys and state under your passphrase, on your hardware | Proven. Ed25519 identity, Argon2id unlock, per-purpose derived subkeys, encrypted state at rest. The passphrase lives in the macOS Keychain or Linux Secret Service and generation fails closed without it. |
| The gate. Policy you set, frozen at startup, the agent cannot read or infer; three approval tiers with a human on the irreversible ones | Proven. Cooperative gates on every MCP call Sanctuary executes or forwards. The wall is what makes the gate more than a request. |
| The record. Append-only, hash-chained audit of what the agent did | Partial. The chain is live; production call sites sign checkpoints when the fortress holds an identity. Fortresses with no identity and the root daemon split audit store write honest unsigned checkpoints. The row stays partial pending the IC-05 external-verifier drill and the remaining local tamper-evidence follow-up. |
| Exit. Keys, state, memory and reputation leave with you | Partial. Export and import ship; three exit-bundle gaps are open (IC-07, IC-08, IC-09). |
Why this shape
Plenty of tools can sandbox an agent when the agent, or its harness, chooses to run inside one. Sanctuary is built for the harder promise: the wall is imposed by the operator, and does not depend on the agent's cooperation. The problem underneath is ambient authority, the fact that code can act with every permission its process has. Language-level answers (capabilities, information-flow typing, functional runtimes) are real and we welcome them, and every one of them still needs something outside the interpreter that holds the credentials, checks the policy and decides what actually leaves the machine. Sanctuary is that something, for whatever runtime you choose. It composes with Concordia (agent negotiation) and Verascore (portable reputation), each in its own repo and neither required.
Why this exists: The Base Layer.
Sanctuary and your harness's sandbox
Your harness already has a sandbox, and the two are built for different questions. The harness sandbox bounds what the agent's shell commands can touch; Sanctuary bounds where the agent account's traffic can go and holds the keys and the record, whether or not the agent cooperates. The middle column states only what each vendor's own page says; the first column states only what an Assurance Matrix row proves.
| Where you run | What Sanctuary adds (matrix row) | What the harness sandbox gives that Sanctuary never claims | Run both? |
|---|---|---|---|
| macOS, wall armed | OS-level egress enforcement below the agent: a signed operator policy with a clean per-uid allow/deny demonstration that survives attended reboot cycles ("Egress enforcement: macOS", proven; one host and one OS version; not an audited per-rule, per-flow trail). Plus custody ("State encryption", proven), approval gates ("Approval gating", proven), and the hash-chained record ("Tamper-evident audit chain", partial) | Filesystem write limits and process-tree limits on the shell commands the harness runs; a network allowlist the harness itself enforces | Yes. The sandbox is chosen by the harness and can be retried outside of; the wall is imposed by the operator below the agent |
| macOS, wall not armed | Custody, approval gates, and the record, as above; no egress enforcement | As above | Yes. Without the wall, egress is bounded only by the harness sandbox's own domain list |
| Ubuntu 24.04 (x86-64) | Kernel egress enforcement per user account on a fresh install of the separate Castle Wall Linux 0.1.0-1 package (a GitHub pre-release, not part of the npm package) ("Egress enforcement: Linux", proven for exactly the claim in the table above and nothing wider). Plus custody, gates, and the record; generated custody needs an unlocked Secret Service | As above | Yes |
| Other Linux distributions or architectures | Custody, gates, and the record; egress enforcement unproven (the matrix row is Ubuntu 24.04 x86-64 only) | As above | Yes, and treat the wall as absent |
| Windows | Read-only custody may open an already-enveloped fortress; custody mutation and install readiness fail closed; enforcement roadmapped | Claude Code: unsandboxed on native Windows, sandboxed under WSL2. Codex: a native Windows sandbox in PowerShell and the Linux sandbox under WSL2. Cursor: no Windows sandbox is described | Nothing to arm on the Sanctuary side yet |
Harness column sources, each fetched 2026-10-12 (none of the three pages shows a last-updated date; the Cursor page's newest changelog entry is dated 2026-10-01):
- Claude Code: Configure the sandboxed Bash tool. The sandbox covers shell commands only and is off by default; it uses Seatbelt on macOS and
bubblewrapplussocaton Linux and WSL2; network goes through a local proxy checked against an allowed-domain list; on native Windows commands run unsandboxed; MCP servers, hooks, and the built-in file and web tools run outside it; Claude can ask to retry a failed command unsandboxed. - Codex: Sandboxing. The sandbox constrains which files the agent can modify and whether commands can use the network; Seatbelt on macOS,
bubblewrapon Linux and WSL2, a native Windows sandbox in PowerShell;workspace-writeis the default mode anddanger-full-accessremoves the filesystem and network boundaries. - Cursor: Run modes. Sandboxed commands read and write inside the workspace; network is blocked by default and then opened per
sandbox.json; Seatbelt throughsandbox-execon macOS, Landlock plus seccomp on Linux kernel 6.2 or later with a Bubblewrap fallback; no Windows sandbox is described; a command that fails on a sandbox restriction can be rerun outside it after classifier review.
The argument for why the two layers are different problems: Containment is an isolation problem.
Start here if you read code
In this order, about an hour:
- ASSURANCE_MATRIX.md. Every claim, its platform, its evidence, its gap, and the next proof owed. A row reads
provenonly on a captured drill; a green suite is not a claim. - SANCTUARY_ARCHITECTURE.md. Entry points, data flow, trust model, and the sovereignty properties stated as testable assertions.
- server/src/README.md. The module map: 64 modules and 56 barrels (counted 2026-10-13), what each owns, and the names that collide.
- The macOS wall drills the matrix cites: allow/deny, [boot su
只显示了开头部分；其余内容请在仓库中阅读