Cron Mcp

io.github.takeaseatventure由 GitHub 账号 takeaseatventure 发布v0.1.2更新于 Oct 25, 2026

MCP cron validator for AI agents; flags impossible schedules and other silent cron bugs.

发布方自己的服务：Croncool Cron Jobs

已验证STDIO仅桌面效率与工作流开发者工具

服务详情

标识符: io.github.takeaseatventure/cron-mcp
传输: STDIO
运行时: Desktop only
信任: Verified
发布于: Sep 16, 2026

安全提示

MCP 服务会接收对话中的工具调用请求，并可能执行外部操作。启用前请先了解该服务能做什么，并只授予它所需的权限。

注册中心已验证此条目所属的 GitHub 账号 takeaseatventure。这并不表明该账号就是产品方。

SourceWeft 的检查只涵盖条目是否完整、端点是否响应；不审查服务器的代码，也不审查其工具的行为。

在 README 中添加徽章

向读者展示该 MCP 服务已收录于 SourceWeft，徽章链接到本页。

[![Listed on SourceWeft](https://sourceweft.com/badge/mcp/io.github.takeaseatventure%2Fcron-mcp.svg)](https://sourceweft.com/mcp/io.github.takeaseatventure%2Fcron-mcp?utm_source=readme&utm_medium=badge)

概览

AI 生成的概览

一个本地 MCP 服务器，让 AI 助手解析、校验、解释和预览 cron 表达式，在部署前标出不可能或高风险的调度。

可以做什么: 把 cron 表达式解析为说明其触发时间的通俗英文描述，支持 5 字段标准 cron 以及 L、W、# 和月份/星期名称写法。深度校验调度，检查不可能日期、日与星期的 OR 语义、午夜高峰、不均匀步长和闰年边界，并估算触发频率。计算接下来 N 次触发时间，以 ISO-8601 时间戳和相对偏移给出。查询常见且经过验证的预设调度库，例如每 5 分钟、每小时、工作日 9 点、每月和每季度。
适用场景: 适合在编写或审查定时任务、CI 流水线或计划任务的 cron 调度时使用，让助手在任务部署前发现永不触发、日与星期 OR 陷阱等隐蔽错误。也适合解释已有表达式或生成下次触发预览。
运行要求: 作为本地 stdio 进程在用户机器上运行，通过 npm 包 cron-mcp-server 安装（全局安装或 npx），需要 Node.js。未声明认证、环境变量或请求头。需在 Claude Desktop、Cursor 或 VS Code Copilot 等桌面客户端中配置启动。

安装前请注意

这些工具只对用户提供的表达式做只读分析；README 未描述写入、发送或向第三方传输数据。它只是对内置 cron 引擎的薄封装，结果取决于该引擎对 cron 语义的解释。不要求任何凭据。

安装

在 SourceWeft 中

打开控制台中的 Cron Mcp，将其添加到工作区。
为需要使用其工具的对话启用该服务。

Desktop only，通过 STDIO。 STDIO 服务会启动本地进程，因此需要 SourceWeft 桌面宿主。

其他 MCP 客户端

参照仓库中的启动说明。

README

cron-mcp

An MCP (Model Context Protocol) server that lets AI coding agents — Claude, Cursor, Copilot, Cline — parse, validate, explain, and preview cron expressions.

Built to catch silent cron bugs (impossible schedules, OR-semantics gotchas, midnight spikes) before a job is deployed — a class of mistake that's easy to make and hard to notice until a critical job silently fails to fire.

Why

Cron expressions are deceptively tricky. Common silent failures:

0 0 30 2 * → never runs (February has no 30th). Syntactically valid, semantically dead.
0 0 1,15 * 1 → fires on the 1st OR 15th OR Monday, not "the 1st and 15th if Monday" (the classic dom+dow OR-semantics trap).
*/7 * * * * → uneven step; intervals drift (:00, :07, :14, …, :56, :00 — not "every 7 minutes" cleanly).
0 0 * * * → midnight spike; every job in the system competes at 00:00.

cron-mcp surfaces these as warnings, observations, and suggestions that the AI agent can act on — before you ship the schedule.

Tools exposed

Tool	Description
`parse_cron`	Parse a cron expression → plain-English description of when it fires. Supports 5-field standard cron + `L` (last), `W` (nearest weekday), `#` (nth weekday), named months/days.
`validate_cron`	Deep validation: impossible schedules, OR-semantics, midnight spikes, uneven steps, leap-year edges, frequency estimate (~runs/year).
`next_runs`	Compute the next N fire times as ISO-8601 + relative offsets.
`cron_presets`	Library of common, proven schedules (every 5 min, hourly, weekdays 9am, monthly, quarterly, …).

Install

Published links

npm package — installable release 0.1.2
Official MCP Registry — registry record
GitHub release — downloadable source checkpoint

Configure

Claude Desktop

Add to ~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or the equivalent on your OS:

Cursor

Add to .cursor/mcp.json (or ~/.cursor/mcp.json for global):

VS Code (Copilot)

Direct (no install)

Example usage

Once connected, just ask the agent in natural language:

"Validate this cron: 0 0 30 2 *"
"When does */5 * * * * next fire?"
"Give me a cron for every weekday at 9am"
"Is there anything risky about 0 0 1,15 * 1?"

The agent calls the tools and returns structured, actionable results.

Engine

The cron engine is battle-tested: 638 lines, zero dependencies, originally extracted from a browser-based cron generator and hardened with 69 unit tests. This MCP server is a thin tool wrapper around it.