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)
Read-only check that a machine meets a project's declared tool and runtime versions.
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 …”。
打开 Claude Desktop → Settings → Developer → Edit Config,把下面的内容粘贴到 claude_desktop_config.json
{
"mcpServers": {
"devready": {
"command": "uvx",
"args": [
"devready-cli",
"mcp"
]
}
}
}在终端中运行这条命令
claude mcp add devready -- uvx devready-cli mcp粘贴到项目中的 .cursor/mcp.json,或 ~/.cursor/mcp.json
{
"mcpServers": {
"devready": {
"command": "uvx",
"args": [
"devready-cli",
"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": {
"devready": {
"type": "stdio",
"command": "uvx",
"args": [
"devready-cli",
"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。
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。
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
注册名称已通过 GitHub 账户 Evalogical-Inc 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
它在你的电脑上运行,没有互联网端点
在你的电脑上运行,改用本地设置
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 1 项适用的基础检查中通过了 1 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)。
GitHub 仓库最后更新于–,最新版本(1.3.0)于前天进入注册中心。
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 Python (uv)。
同一分类中有:Spec Driven Development, Remote Desktop Commander, zanadir
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.
Find what a repository's CI is missing, and which security checks cannot fail the build.
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。
A zero-dependency, read-only CLI that checks whether your machine actually meets a project's declared requirements, and tells you exactly what to fix. Built for humans, CI, and AI coding agents (Claude Code, Codex, any MCP client).
One command, every requirement checked, exact fixes shown.
DevReady is a command-line tool that tells you whether your computer is set up to work on a project. Run devready check inside a project folder and it:
.nvmrc, .python-version, package.json engines, pyproject.toml requires-python).1 if a requirement is unmet, so it also works in CI.It only looks. It never installs, changes or deletes anything, makes no network calls and sends no telemetry. Fixing the problem (for example switching Node versions) is up to you.
You clone a repo. npm install explodes. The README says "Node 20", your machine has 16, and the error message says nothing useful. Multiply by every new hire, student, CI runner and AI agent that burns tokens guessing why a build failed.
$ devready check
DevReady - Environment Check
[PASS ] Python 3.15.0 found [PASS ] Git 2.52.0 found [PASS ] Node.js 23.5.0 found [PASS ] npm 11.14.1 found [FAIL ] Node engines 23.5.0 does not satisfy package.json engines (>=99) [WARN ] Python pin 3.15.0 does not satisfy .python-version (3.99)
4 passed, 1 warning, 1 failed No changes were made to your system.
Output above is from a real run (pins deliberately set to impossible values to show failures). Exit code
1when a requirement is unmet, so it drops straight into CI.
| 🎯 Project-aware | Reads what *your project* declares (.nvmrc, .node-version, .python-version, package.json engines, pyproject.toml requires-python) and compares it to what is installed. Not just "is Node installed?" |
| 🔒 Read-only by design | Never installs, edits or deletes. Never prints env values. No network. No telemetry. |
| 🤖 Agent-ready | Stable, versioned JSON schema, a Claude Code plugin, an MCP server and AGENTS.md. Agents check first instead of guessing. |
| 🛡️ Prompt-injection safe | Output contains only parsed versions and fixed text, never raw file or command output, so a hostile package.json can't smuggle instructions to your agent (tested). |
| 📦 Zero dependencies | Pure Python standard library. Installs in a second, runs anywhere Python 3.10+ does. |
| ⚙️ CI-friendly | Documented exit codes, JSON output, tested on Linux, macOS and Windows in CI. |
devready check in your README's setup steps and cut "it doesn't work" issues.Install from PyPI:
pipx install devready-cli
uvx devready-cli check
From npm (runs the same PyPI release through uvx or pipx, so one of them must be installed):
npx devready-cli check
No Python tooling yet (for example a JavaScript-only machine)? uv installs without Python, and uvx then runs DevReady with no further setup:
# macOS/Linux
curl -LsSf https://astral.sh/uv/install.sh | sh
powershell -ExecutionPolicy ByPass -c "irm https://astral.sh/uv/install.ps1 | iex"
uvx devready-cli check
From a clone (for development):
git clone https://github.com/Evalogical-Inc/devReady && cd devReady
python -m venv .venv
python -m pip install -e ".[dev]"
devready check # human-readable report for the current directory
devready check --quiet # only non-pass results (text format) devready check --project ../app # inspect another project's pins devready check --format json # machine-readable (see docs/schema.json) devready explain docker.daemon # what a check means and how to fix it devready check --format junit # or sarif, for CI devready check --only project --skip project.ports # filter by id or prefix devready fix --plan # suggested steps only; never runs them devready check --format json > baseline.json # save a snapshot (commit it for a team baseline) devready check --baseline baseline.json # show only what changed; fail only on new/worse failures devready init > .devready.json # optional config: skip ids, require node/python, required env names devready mcp # run as an MCP server over stdio devready --version
| Exit code | Meaning |
|---|---|
0 | No blocking failures |
1 | A project requirement is not met |
2 | Invalid usage (e.g. bad --project path) |
3 | Internal error |
| Status | Meaning |
|---|---|
pass | Requirement met |
warn | Tool missing, or a soft pin (.nvmrc, .python-version) differs |
fail | A hard requirement (engines, requires-python) is not met |
skip | Not evaluated (tool absent, or requirement syntax not understood) |
error | DevReady couldn't run the check |
{
"schema_version": 1, "results": [ { "check_id": "project.node.engines", "title": "Node engines", "status": "fail", "summary": "23.5.0 does not satisfy package.json engines (>=99)", "suggestion": "Switch versions with your version manager (nvm, pyenv, mise, ...) or update the requirement.", "docs_url": "" } ] }
- uses: Evalogical-Inc/devReady@v1.3.0
with: skip: project.ports # optional; also: only, format, project, version
To show findings in GitHub code scanning, save a SARIF report and upload it. Pin findings point at the file that declares the pin (package.json, .nvmrc, go.mod, ...); tool-only findings have no file.
- uses: Evalogical-Inc/devReady@v1.3.0
with: format: sarif output: devready.sarif continue-on-error: true # still upload when a requirement is unmet
The job fails when a declared requirement is unmet (fail). It installs devready-cli from PyPI, so flags added after your installed release (--only, --skip, junit/sarif) need a release that includes them; set version: to pin one.
Tell contributors your project declares its requirements and can be checked with one command. Paste this into your README:
[](https://github.com/Evalogical-Inc/devReady)
DevReady ships the same engine three ways, so your agent can verify the environment *before* it installs, builds or tests.
Claude Code - plugin with a skill, a /devready:doctor command and an MCP server (the MCP server needs uv; the rest works without it once devready-cli is installed):
/plugin marketplace add Evalogical-Inc/devReady
/plugin install devready --marketplace Evalogical-Inc/devReady
Any MCP client (Codex, Cursor, ...) - one read-only tool, check_environment. Also listed in the MCP Registry as io.github.Evalogical-Inc/devready:
# ~/.codex/config.toml (verify keys against your Codex version)
[mcp_servers.devready] command = "uvx" args = ["--from", "devready-cli", "devready", "mcp"]
Anything else - drop AGENTS.md in your repo, or copy skills/devready. Details in docs/agents.md.
See docs/security.md for the model, docs/versioning.md for what stays stable, docs/checks.md for every check, and SECURITY.md to report a vulnerability.
.nvmrc, .node-version, .python-version, .tool-versions, package.json engines, pyproject.toml requires-python), .env.example vs .env key names, Docker CLI + daemon, compose file validity and image platform (arm64 vs amd64) mismatch (only when the project has a Dockerfile/compose file), free disk space, Windows long paths, PowerShell policy, line endings, WSL version and Docker Desktop engine, macOS/Linux build tools, and (only when the project declares them) Go, Ruby, Java, Rust, .NET, pnpm, Yarn and Bun pins, .env not git-ignored, node_modules vs lockfile, and declared ports already in use.skip rather than guess..venv/venv, else the first python on PATH), and the result names the interpreter checked. DevReady does not look at other virtualenv locations (poetry, conda, pyenv shims)..tool-versions, .env key-name diff, disk space, git identity, devready explainfix --plan, GitHub Action.devready.json config, --baseline snapshotsfnm)--recursive, org `--polic只显示了开头部分;其余内容请在仓库中阅读