能做什么
Scans code for hardcoded secrets (AWS, Stripe, GitHub, JWTs) before an AI agent commits it.
作者提供的原文
- 连接方式 在你的电脑上运行
- 账号或密钥 无需密钥
- 分类 开发者工具
可搭配使用
教 AI 做这类工作的 Skill,可与本项一起使用
andrej-karpathy-skills
A single CLAUDE.md file to improve Claude Code behavior, derived from Andrej Karpathy's observations on LLM coding pitfalls.
graphify
Turn any codebase, with its docs, SQL schemas, configs, and PDFs, into a queryable knowledge graph. A /graphify skill for Claude Code, Cursor, Codex, and Gemini CLI: local deterministic AST parsing, every edge explained, no vector store.
为什么排在这里
列表按使用分(满分 100)排序,由以下部分组成
| 热度 | 0/80 | GitHub 星标或每周下载量:较大的计 60,较小的计 20 |
|---|---|---|
| 新鲜度 | 3/12 | 30 天内有发布或代码更新得满分 |
| 仍在响应 | 0/8 | 上次检查时云端端点有响应 |
| 使用分 | 3/100 |
质量: 基础检查通过 3/3 项(仅展示,不用于排序) · 上升势头: 暂无数据 (只用于“上升中”标签)
安装
选择你使用的 AI 应用,然后复制设置。
通过 3/3 项检查 · 在你的电脑上运行
第一步:你用哪个 AI?
选择你用的应用,步骤会显示在下方。
可连接 Claude,但需要先准备
不需要密钥,不用安装程序 · 仅限电脑上的 Claude 应用(Windows 或 Mac),网页和手机不可用
双击文件,然后在电脑版 Claude 应用中点击 Install
试一试
用 Secretguard MCP 告诉我你能帮我做什么,然后做一件给我看✓ 如果 Claude 说正在使用 Secretguard MCP,或回答中出现 Secretguard MCP 的工具名称,就表示连接成功
菜单名称已于 2026年11月11日 对照 Anthropic 帮助页面核实 · 你的界面可能因版本略有不同
ChatGPT 暂时无法连接 Secretguard MCP
Secretguard MCP 需要安装在你的电脑上,而 ChatGPT 只能连接在网上运行的服务器
你可以改用这些方式
A · 使用能连接 Secretguard MCP 的应用
- 不需要密钥,不用安装程序 · 仅限电脑上的 Claude 应用(Windows 或 Mac),网页和手机不可用
B · 开发者工具 类中能连接 ChatGPT 的其他服务器
从同类中使用较多的里挑选,仅限用地址即可连接、无需密钥的
- Sentry✓ ChatGPT
- Compiler Explorer✓ ChatGPT
- GitDiagram✓ ChatGPT
Claude Code 暂时无法连接 Secretguard MCP
开发者只提供了电脑版 Claude 应用的安装文件
你可以改用这些方式
A · 使用能连接 Secretguard MCP 的应用
- 不需要密钥,不用安装程序 · 仅限电脑上的 Claude 应用(Windows 或 Mac),网页和手机不可用
Cursor 暂时无法连接 Secretguard MCP
开发者只提供了电脑版 Claude 应用的安装文件
你可以改用这些方式
A · 使用能连接 Secretguard MCP 的应用
- 不需要密钥,不用安装程序 · 仅限电脑上的 Claude 应用(Windows 或 Mac),网页和手机不可用
VS Code 暂时无法连接 Secretguard MCP
开发者只提供了电脑版 Claude 应用的安装文件
你可以改用这些方式
A · 使用能连接 Secretguard MCP 的应用
- 不需要密钥,不用安装程序 · 仅限电脑上的 Claude 应用(Windows 或 Mac),网页和手机不可用
一览
哪些应用可以使用它
依据这个服务器的连接方式和各应用自己的文档,检查于 2026年11月10日。
在你的电脑上运行,适用于
ClaudeGooseLM StudioMsty StudioCherry StudioVisual StudioWindsurf (Devin Desktop)ZedKiroClineContinueJetBrains AI AssistantCodex CLIGemini CLIOpenCodenodecodeWarpLibreChatRaycast
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20日
- ✓ 通过
作者与仓库一致
注册名称已通过 GitHub 账户 vladimirbakalov 验证,该账户拥有此仓库
- ✓ 通过
仍在维护
代码最后更新于3个月前
- ✓ 通过
有许可证
使用 MIT
- – 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
作者:申请重新检查或报告信息错误
给作者:在 README 中加徽章
如果你是作者,可把这段代码贴到 README。徽章会链接回本页。
[](https://mcp.popnx.com/en/servers/vladimirbakalov-secretguard-mcp)讨论
提问、分享技巧或报告无法使用。按“有帮助”数量排序。
用户评价
还没有评价
与“开发者工具”中的其他服务器比较i
- 星标数多于此分类中其他 6209 个服务器的 0%
- 发布到注册中心的时间比此分类中其他 6209 个服务器的 25% 更新
- 基础检查结果等于或优于此分类中其他 6209 个服务器的 99%
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
Secretguard MCP 安全吗?
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。
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3个月前,最新版本(0.1.4)于3个月前进入注册中心。
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 MCPB。
有什么可以替代它?
同一分类中有:GitDiagram, Appllama, Cloudflare MCP, Omnara, Repowise
功能相近的其他选择
同一分类、其他作者、命令已通过检查。官方的排在前面,再按星标排序。
如何看卡片:↓ 每周下载量i★ 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
GitDiagram
Architecture diagrams, explanations and Mermaid source for public GitHub repositories.
Appllama
Study screens, flows and paywalls from top-earning iOS apps, then build from what wins.
Omnara
Deploy durable AI agents with Omnara, an open-source managed-agent platform.
Repowise
Codebase intelligence for AI coding agents: graph, git history, docs, decisions, code health.
| 本项 Secretguard MCP | GitDiagram | Appllama | Cloudflare MCP | Omnara | Repowise | |
|---|---|---|---|---|---|---|
| 运行位置 | 需要安装在电脑上 | 在云端运行,无需安装 | 在云端运行,无需安装 | 在云端运行,无需安装 | 在云端运行,无需安装 | 云端或安装在电脑上都可以 |
| 需要密钥 | 否 | 否 | 否 | 否 | 否 | 否 |
| 可在 Claude 中按 + 添加 | 否 | 否 | 否 | 是 | 否 | 否 |
| 基础检查 | 3/3 | 4/4 | 4/4 | 4/4 | 4/4 | 5/5 |
| 许可证 | MIT | MIT | MIT | Apache-2.0 | Apache-2.0 | AGPL-3.0 |
| 价格 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 | 登记中未列出费用 |
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。
secretguard-mcp
An MCP (Model Context Protocol) server that scans a code string for hardcoded secrets — AWS keys, Stripe keys, GitHub tokens, Google API keys and OAuth client secrets, Slack tokens and incoming webhook URLs, Shopify access tokens, Telegram bot tokens, DigitalOcean tokens, Hugging Face tokens, Notion API tokens, Mailchimp API keys, Postman API tokens, Linear API keys, Readme API keys, Clojars API tokens, Pulumi API tokens, OpenAI keys, Anthropic keys, npm access tokens, SendGrid keys, Twilio API keys, Azure Storage account keys, database connection strings with embedded passwords, private key blocks, JWTs, and generic high-entropy credentials — so an AI coding agent (Claude Code, Cursor, Windsurf, ...) can catch a secret *before* it writes the file or makes the commit, instead of finding out at CI/PR-review time. It exposes exactly one tool, scan_for_secrets, runs entirely locally over stdio, needs no API key, and never returns a raw secret value — every finding comes back redacted.
Why this exists
secret-scan-action already catches these secrets in CI, on every PR. That's necessary but late — by the time it runs, the secret has already been written, committed, and pushed. This project reuses that same detection engine (same rules, same entropy check, same redaction) but puts it in front of the agent as a tool call, so the check can happen at generation time, before the secret ever touches disk or history.
What it does
On a scan_for_secrets call:
- Splits the input
codestring into lines. - Runs the same two-tier ruleset
secret-scan-actionuses:
- Pattern rules (high confidence) — distinctive formats that are near-certain secrets when matched: AWS access key IDs (
AKIA...) and contextual secret keys, Stripe live keys (sk_live_,rk_live_), GitHub tokens (ghp_,gho_,github_pat_, ...), Google API keys (AIza...), Google OAuth client secrets (GOCSPX-...), Slack tokens (xox[baprs]-...), Slack incoming webhook URLs (hooks.slack.com/services/...), Shopify access tokens (shpat_...,shpca_...,shpss_...,shppa_...,shpua_...), Telegram bot tokens (:A..., 35-char secret), DigitalOcean tokens (dop_v1_...,doo_v1_...,dor_v1_..., 64-char hex), Hugging Face tokens (hf_...,api_org_..., 34-char alpha), Notion API tokens (ntn_..., 11 digits + 35 alphanumeric), OpenAI keys (sk-...,sk-proj-...,sk-svcacct-...), Anthropic keys (sk-ant-...), npm access tokens (npm_...), SendGrid keys (SG....), Twilio API keys (SK...), Azure Storage account keys (contextualAccountKey=...), private key blocks (-----BEGIN ... PRIVATE KEY-----), and JWTs. One pattern rule — database connection strings with an embedded password (postgres://,mysql://,mongodb(+srv)://,redis(s)://,amqp(s)://) — is deliberately *not* near-certain even after excluding known placeholder passwords (user,password,changeit, ...) and${...}-style env-var references, since a real value there could still be a low-stakes tutorial example rather than a live credential; it's returned at generic confidence, same as the entropy rule below. Another pattern rule — Mailchimp API keys (a 32-char hex value followed by a-usNNdatacenter suffix) — is also generic confidence: it only fires when amailchimp-prefixed variable/key name immediately precedes the value, but that keyword gate still doesn't rule out an unrelated hex value that happens to end in the same suffix shape. Postman API tokens (PMAK-..., 24-char hex +-+ 34-char hex), Linear API keys (lin_api_..., 40-char alphanumeric), Readme API keys (rdme_..., 70-char lowercase alphanumeric), Clojars API tokens (CLOJARS_..., case-insensitive, 60-char alphanumeric), and Pulumi API tokens (pul-..., 40-char lowercase hex) are high confidence — a fixed prefix and exact length, same as the other provider-token rules. - Generic entropy rule — a value assigned to a variable named like
secret,token,password/credential, or a*keycompound commonly used for real secret material (apiKey,sessionKey,signingKey,clientKey,webhookKey, ...) whose value also has high Shannon entropy (looks random, not like a placeholder or an env-var reference). Deliberately does *not* match a bare*Key— that would also catchpartitionKey,cacheKey,queryKey, and similar non-secret identifiers common in ordinary code.
- Returns every finding's
filename,line,ruleId,description,confidence("high"|"generic"), and a redacted line — the raw secret value never leaves the process. If nothing is found, it returns a plain "No secrets detected." result.
Example output
Calling scan_for_secrets with:
{
"code": "const key = \"[PopNx MCP ตัดค่าที่หน้าตาเหมือนกุญแจออก]\";\nconst greeting = \"hello\";", "filename": "src/config.ts" }
returns:
{
"findings": [ { "filename": "src/config.ts", "line": 1, "ruleId": "aws-access-key-id", "description": "AWS Access Key ID", "confidence": "high", "redactedLine": "const key = \"AKIA**********MPLE\";" } ], "summary": "Found 1 potential secret (1 high-confidence, 0 needs-review).\n\n- [high] src/config.ts:1 — AWS Access Key ID (aws-access-key-id)\n const key = \"AKIA**********MPLE\";" }
(The AWS key above is AWS's own public documentation placeholder, not a live credential.) A clean scan — e.g. { "code": "const greeting = \"hello world\";" } — returns { "findings": [], "summary": "No secrets detected." }.
Setup
Not yet published to the npm registry — install directly from GitHub via npx. npm install from a git source runs this package's prepare script automatically, which builds dist/ on the fly, so no separate build step is needed.
Claude Code
Add to your project's .mcp.json (or run claude mcp add):
{
"mcpServers": { "secretguard": { "command": "npx", "args": ["-y", "github:vladimirbakalov/secretguard-mcp"] } } }
Claude Desktop
Add to claude_desktop_config.json:
{
"mcpServers": { "secretguard": { "command": "npx", "args": ["-y", "github:vladimirbakalov/secretguard-mcp"] } } }
No API key, no account, no config options — restart Claude Code / Claude Desktop and scan_for_secrets is available. The tool description tells the agent to call it before writing code that could contain a credential, and again before a commit or PR — most of the time you won't need to ask for it explicitly.
Cursor / Windsurf
Both read the same command/args shape from their own MCP settings UI or config file — point them at npx -y github:vladimirbakalov/secretguard-mcp the same way.
Once this package is published to npm, the args above can drop to ["-y", "secretguard-mcp"] instead — that's a follow-up, not a blocker.
One-click install (.mcpb)
A prebuilt MCP Bundle is attached to the v0.1.4-mcpb release — download secretguard-mcp-0.1.4.mcpb and open it in Claude Desktop (or any other MCPB-compatible client) for a one-click local install, no npx/Node setup required on the client side. Rebuild it yourself with npm run package:mcpb (see scripts/build-mcpb.sh).
This same .mcpb release asset is what server.json at the repo root points at for the official MCP Registry — secretguard-mcp is published and listed there as io.github.vladimirbakalov/secretguard-mcp, so MCP clients that browse the official registry can discover and install it directly, in addition to the npx/.mcpb paths above. Publishing runs unattended in CI (.github/workflows/publish-mcp.yml) via mcp-publisher login github-oidc on every v*-mcpb tag push — no interactive login step.
Security notes
- The raw secret value matched by a rule is held in memory only for the duration of a single
scan_for_secretscall and is redacted (redactLine/redactSecret) before the tool result is built — it never appears in the returnedcontent,structuredContent, or any log line. - The server does no network calls of any kind. It reads stdin, writes stdout (MCP stdio transport), and does nothing else.
- Generic-tier findings are ambiguous by nature (config placeholders, hashes, and UUIDs can trip the entropy check) — that's expected. Treat
confidence: "generic"as "worth a second look," not "confirmed."
Development
npm install
npm run typecheck # tsc --noEmit npm test # vitest run npm run build # tsc -p tsconfig.build.json -> dist/
dist/ is not committed — it's built from src/ via the prepare script, which runs both on a git-based npx/npm install and before any future npm publish.
Scope (v1)
One tool, one job: scan a code string, return redacted findings. No allowlist file, no AI triage step, no config options, no persistent state. If this needs any of that later, it'll get added once real usage shows it's needed — not before.
Relationship to secret-scan-action
secretguard-mcp and secret-scan-action share the same detection engine (rules.ts, redact.ts, and the core of scan.ts) but are independent, separately distributed packages: one is a GitHub Action that scans PR diffs in CI, the other is an MCP server that scans arbitrary code strings locally, before a commit exists. Fixing a false positive/negative in the ruleset means updating both.
License
MIT.