Permet aux agents IA de lister les tâches planifiées, de lire les journaux d'erreurs, d'exécuter des tests de santé et de mettre en sourdine les alertes pour les tâches cron surveillées.
Aussi connu sous cronwatchdev/cronwatch, @cronwatch/sdk
- Étoiles
- 26
- Outils
- —
- Téléchargements, 30 jours
- 1 127
- Dernière version
- —
- Dernier commit
- 14 nov. 2026
- Licence
- MIT
Qu'est-ce que CronWatch MCP ?
Permet aux agents IA de lister les tâches planifiées, de lire les journaux d'erreurs, d'exécuter des tests de santé et de mettre en sourdine les alertes pour les tâches cron surveillées.
Résumé généré à partir du README du projet
À propos du MCP server CronWatch
CronWatch permet de surveiller les tâches planifiées et d'enquêter sur les échecs directement depuis un agent IA. Ce MCP server permet à des outils comme Claude Code et Cursor d'interagir directement avec les tâches planifiées de votre application. Il offre aux agents la capacité de lister les tâches actives, de lire les logs des exécutions ayant échoué, d'exécuter des tests de santé et de mettre en sourdine les alertes actives.
Les développeurs l'intègrent en ajoutant le SDK cronwatch à leurs applications existantes dans des langages tels que TypeScript, Python, Ruby, Go, Rust, PHP, Elixir, Java ou .NET. Le système utilise votre base de données actuelle, comme SQLite, Postgres ou MySQL, pour enregistrer les exécutions de tâches sans nécessiter de serveur séparé ou de compte externe. Une fois le SDK implémenté dans votre code, le MCP server connecte votre client IA à vos données de tâches internes, permettant un triage et un débogage automatisés des tâches manquées, lentes ou ayant échoué.
Ce que vous pouvez faire avec
- 1Auditer les calendriers de tâches activesDemandez une liste complète de toutes les tâches cron actuellement surveillées au sein de votre application.
- 2Gérer les notifications d'alerteDemandez à votre assistant IA de mettre en sourdine les alertes pour une tâche spécifique pendant une fenêtre de maintenance.
- 3Effectuer des tests de santé du systèmeDemandez à un agent de déclencher un test de santé pour s'assurer que le système de surveillance cron fonctionne correctement.
Extrait du README du projet. Chaque cas d'utilisation est vérifié par rapport à une citation du document.
Comment installer CronWatch MCP
Formats vérifiés le 10 nov. 2026Ajouter à CursorAjouter à GooseInstaller CronWatch MCP dans Claude Code
$ claude mcp add cronwatch -- npx -y @cronwatch/sdkInstaller CronWatch MCP dans Cursor
~/.cursor/mcp.json
{
"mcpServers": {
"cronwatch": {
"command": "npx",
"args": [
"-y",
"@cronwatch/sdk"
]
}
}
}Installer CronWatch MCP dans VS Code
.vscode/mcp.json
{
"servers": {
"cronwatch": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"@cronwatch/sdk"
]
}
}
}Installer CronWatch MCP dans VS Code CLI
$ code --add-mcp '{"name":"cronwatch","type":"stdio","command":"npx","args":["-y","@cronwatch/sdk"]}'Installer CronWatch MCP dans Claude Desktop
claude_desktop_config.json
{
"mcpServers": {
"cronwatch": {
"command": "npx",
"args": [
"-y",
"@cronwatch/sdk"
]
}
}
}Installer CronWatch MCP dans Codex
$ codex mcp add cronwatch -- npx -y @cronwatch/sdkInstaller CronWatch MCP dans Codex config.toml
~/.codex/config.toml
[mcp_servers.cronwatch] command = "npx" args = ["-y", "@cronwatch/sdk"]
Installer CronWatch MCP dans Gemini CLI
$ gemini mcp add -s user cronwatch npx -- -y @cronwatch/sdkInstaller CronWatch MCP dans Gemini settings.json
~/.gemini/settings.json
{
"mcpServers": {
"cronwatch": {
"command": "npx",
"args": [
"-y",
"@cronwatch/sdk"
]
}
}
}Vérifications automatisées
Exécuté le 13 nov. 2026Rapport complet →A passé nos vérifications automatisées. Il ne s'agit pas d'un audit de sécurité : lisez le code et donnez au serveur l'accès le plus restreint dont il a besoin. Fonctionnement des vérifications
- Dépôt source lié et accessibleRéussi
- Licence open source (OSI)Réussi
- Commit ou version publiée au cours de l'année écouléeRéussi
- Nom du package non proche d'un package populaireRéussi
- L'endpoint distant utilise HTTPSNon vérifié
- Les outils destructifs sont étiquetésNon vérifié
- Aucune instruction cachée destinée au modèleRéussi
- Aucun avis critique dans le package ou ses dépendancesNon vérifié
README
Voir sur GitHubCronWatch
Cron and scheduled-job monitoring that lives inside your app. Wrap a job once; CronWatch records every run in a database you already have and tells you when a run is missed, fails, gets stuck, runs slow, goes over budget, or quietly does nothing. No server to run, no account to make. MIT.
Site and docs: cronwatch.dev
// lib/cronwatch.ts
import { cronwatch } from "@cronwatch/sdk";
import { sqlite } from "@cronwatch/sdk/sqlite";
import { slack } from "@cronwatch/sdk/slack";
export const cw = cronwatch({
store: sqlite({ path: "./data/cronwatch.db" }),
alerts: [slack({ webhookUrl: process.env.SLACK_WEBHOOK_URL! })],
});
export const nightlyReport = cw.job("nightly-report", {
schedule: "0 2 * * *", timezone: "UTC", grace: "15m", expect: "Report written",
});
// app/api/cron/nightly-report/route.ts: checks Authorization: Bearer CRON_SECRET
import { nightlyReport } from "@/lib/cronwatch";
export const GET = nightlyReport.handler(async (job) => {
const report = await buildReport();
job.log("Report written:", report.path);
job.metric("cost", report.usdCost);
});
// app/cronwatch/[[...path]]/route.ts: a dashboard and JSON API behind CRONWATCH_TOKEN
import { cw } from "@/lib/cronwatch";
export const { GET, POST, DELETE } = cw.routes();
Packages
| Package | What |
|---|---|
@cronwatch/sdk |
the library for Node, Cloudflare Workers, Deno, and Bun: jobs, runs, checks, stores (memory, SQLite, Postgres, D1), pg_cron jobs read from Postgres, alerts (Slack, Discord, webhook, email, SMS, error trackers), dashboard and API, adapters for Node servers, optional Claude triage |
cronwatch gem |
the Ruby port for Ruby and Rails apps: ActiveRecord store, ActiveJob and Sidekiq integration, schedules read from Solid Queue or sidekiq-cron, pg_cron jobs, the same alert channels, a check job, the dashboard as a Rack app. Same rules, alerts, and stored rows as the SDK |
cronwatch-sdk for Python |
the Python port for Python apps: the memory, SQLite, and Postgres stores, the same alert channels and Claude triage, pg_cron jobs, the dashboard as a WSGI and ASGI app, Django, Celery and beat, APScheduler, async jobs, and handler() for platform crons and AWS Lambda |
cronwatch/cronwatch for PHP |
the PHP port for PHP apps: the memory, SQLite, MySQL (and MariaDB), and Postgres stores, sharing tables with the SDK byte for byte, the same alert channels and Claude triage, pg_cron jobs, vendor/bin/cronwatch check, the dashboard (a bare script or PSR-15), handler() for platform crons, Laravel and Symfony integrations, a WordPress plugin, a Drupal module (drupal/cronwatch), and a Craft CMS plugin (cronwatch/craft) (its DESIGN.md) |
cronwatch.dev/go |
the Go port for Go apps: jobs, runs, and checks with context throughout and safe across goroutines, the memory store and a database/sql store for SQLite, Postgres, MySQL, and MariaDB over the app's own driver, sharing tables with the SDK byte for byte, the same alert channels and Claude triage, pg_cron jobs, the dashboard and job.Handler as http.Handlers, an AWS Lambda adapter, and modules of their own for robfig/cron, gocron, River, and Asynq (its DESIGN.md) |
cronwatch crate |
the Rust port for Rust services, on tokio: jobs, runs, and checks, safe across tasks, the memory store and cronwatch-sqlx for SQLite, Postgres, MySQL, and MariaDB over the app's own sqlx pool, sharing tables with the SDK byte for byte, the same alert channels and Claude triage, pg_cron jobs, the dashboard and handler() framework-free with tower and axum adapters (and so AWS Lambda through lambda_http), a blocking client, and crates of their own for tokio-cron-scheduler and apalis (its DESIGN.md) |
cronwatch on Hex |
the Elixir port for Elixir and Erlang services: an instance in the app's supervision tree, jobs run in the calling process and recorded when it dies, the memory store and an Ecto store for SQLite, Postgres, MySQL, and MariaDB over the app's own repo, sharing tables with the SDK byte for byte, the same alert channels and Claude triage over OTP's own sockets, pg_cron jobs, the dashboard and a job's handler as Plugs for a Phoenix router, telemetry events, a release's crontab check, and integrations for Oban and Quantum (its DESIGN.md) |
dev.cronwatch:cronwatch for Java |
the Java port for services on the JVM: jobs run in the calling thread and safe across platform and virtual threads, the current run carried across threads, a shutdown hook that records runs a stopping JVM leaves open, the memory store and a JDBC store for SQLite, Postgres, MySQL, and MariaDB over the app's own DataSource, sharing tables with the SDK byte for byte, the same alert channels and Claude triage over the JDK's own HTTP client, pg_cron jobs, the dashboard and a job's handler framework-free with adapters for the JDK's server, servlet containers (cronwatch-servlet), and Spring MVC and WebFlux, CronwatchCli for a crontab's check, a Spring Boot starter that watches every @Scheduled method (ShedLock included), and modules of their own for Quartz and JobRunr (its DESIGN.md) |
Cronwatch on NuGet |
the .NET port for .NET services: jobs run in the caller's flow and safe across threads and tasks, the current run carried across awaits, a process-exit hook that records runs a stopping process leaves open, the memory store and an ADO.NET store for SQLite, Postgres, MySQL, and MariaDB over the app's own DbDataSource, sharing tables with the SDK byte for byte, the same alert channels and Claude triage over HttpClient, pg_cron jobs, the dashboard and a job's handler framework-free with an ASP.NET Core adapter (Cronwatch.AspNetCore), the client in the Generic Host with hosted jobs on a cron (Cronwatch.Hosting), CronwatchCli for a crontab's check, Native AOT support, and packages of their own for Hangfire and Quartz.NET (its DESIGN.md) |
@cronwatch/mcp |
an MCP server so Claude Code, Cursor, and other agents can list jobs, read failures, run a check, and silence alerts |
skills/cronwatch |
an agent skill for Claude Code, Codex, Cursor, and others: how to add monitoring to a job and how to investigate a failure. npx skills add cronwatchdev/cronwatch installs it (on skills.sh) |
Installing
| Language | Install | Docs |
|---|---|---|
| TypeScript and Node | npm install @cronwatch/sdk, and the driver for a store (below) |
Getting started |
| Ruby and Rails | bundle add cronwatch, then bin/rails generate cronwatch:install and bin/rails db:migrate in a Rails app (Ruby 3.2 or newer; Rails 7.2, 8.0, and 8.1) |
Rails, Ruby, its README |
| Python | pip install cronwatch-sdk |
Python, its README |
| PHP | composer require cronwatch/cronwatch |
PHP, its README |
| WordPress | CronWatch in the plugin directory: search for it under Plugins, Add New Plugin, or wp plugin install cronwatch --activate |
WordPress |
| Drupal | composer require drupal/cronwatch, then drush pm:install cronwatch |
Drupal, drupal.org |
| Craft CMS | composer require cronwatch/craft, then php craft plugin/install cronwatch, or from the Plugin Store |
Craft CMS |
| Go | go get cronwatch.dev/go, and go get cronwatch.dev/go/robfigcron (or /gocron, /river, /asynq) for a scheduler |
Go, Go schedulers, its README |
| Rust | cargo add cronwatch --features alerts, cargo add cronwatch-sqlx --features postgres (or sqlite, mysql), cargo add sqlx --no-default-features --features runtime-tokio,postgres, and cargo add tokio --features macros,rt-multi-thread; cronwatch-tokio-cron-scheduler or cronwatch-apalis for a scheduler |
Rust, Rust schedulers, its README |
| Elixir | {:cronwatch, "~> 0.14"} in mix.exs, with the app's Ecto adapter for the SQL store; Oban and Quantum are watched with no other package |
Elixir, Elixir schedulers, its README |
| Java | dev.cronwatch:cronwatch (Java 21 or newer), or dev.cronwatch:cronwatch-spring-boot-starter in a Spring Boot app; cronwatch-servlet, cronwatch-quartz, and cronwatch-jobrunr beside it at the same version, with the app's JDBC driver for the SQL store |
Java, Java schedulers, its README |
| .NET | dotnet add package Cronwatch (.NET 10 or newer), Cronwatch.Hosting in a Generic Host app or Cronwatch.AspNetCore in an ASP.NET Core app; Cronwatch.Hangfire and Cronwatch.Quartz beside it at the same version, with the app's ADO.NET driver for the SQL store |
.NET, .NET schedulers, its README |
The SDK depends only on croner. The core, the D1 store, the pg_cron source, and every alert channel use only fetch and Web Crypto, so they run on Node 22 or newer, Cloudflare Workers, Deno, and Bun; the SQLite and Postgres stores and @cronwatch/sdk/node need Node. Each driver is an optional peer, installed only when you use its entry point:
| Entry point | Install |
|---|---|
@cronwatch/sdk/sqlite |
better-sqlite3 (and @types/better-sqlite3 for TypeScript) |
@cronwatch/sdk/postgres |
pg (and @types/pg for TypeScript) |
@cronwatch/sdk/anthropic |
@anthropic-ai/sdk 0.115 up to (not including) 0.129 |
@cronwatch/sdk/pg-cron |
nothing: it queries through the pg Pool (or anything with query()) you pass it |
@cronwatch/sdk/d1, /node, and the channels (/slack, /discord, /webhook, /resend, /postmark, /sendgrid, /mailgun, /ses, /twilio, /sentry, /honeybadger, /datadog, /rollbar, /bugsnag, /newrelic) |
nothing |
The store entry points' type declarations refer to the driver's types, so a TypeScript project using them without @types/better-sqlite3 or @types/pg fails with TS7016 unless skipLibCheck is on. The package ships ESM and CommonJS, each with its own types.
One library, nine languages
Every language's package follows the same design: each writes the same tables and sends the same alerts, so a Rails app, a Go service, and a Node worker can share one database and one dashboard. All of them are tested against the same recorded cases, so a job behaves the same whichever language runs it. Each package's README and its page on cronwatch.dev have the examples for its language and its schedulers.
Stability
From 1.0 every package follows semantic versioning, one version for all of them. Stability says what a 1.x release promises (the documented API, the stored data, the JSON API, the webhook, settings and environment variable names, documented behaviour) and what it does not; Environment variables lists every variable each library reads; Deprecations lists every deprecated name with its replacement. What changed in each release is in CHANGELOG.md.
Why a library and not a service
A hosted monitor gives you an observer that is alive when your job is not. That is real, and it is the one thing a library cannot do: if your whole app is down, nothing inside it can alert (pair it with any uptime monitor for that case). Everything else, from a job that never fires to one that costs three times what it should, is caught from within, with your run history in your own database and nothing to sign up for.
Contributing
Bugs, ideas, and questions are welcome in the issue tracker. CONTRIBUTING.md covers working on CronWatch itself: each language's tests, releases, and the site.
License
MIT, see LICENSE.
Questions sur CronWatch MCP
Comment ajouter CronWatch MCP à Claude Code ?
Exécutez claude mcp add cronwatch -- npx -y @cronwatch/sdk dans votre terminal. Ajoutez --scope user pour le rendre disponible dans tous les projets.
Comment ajouter CronWatch MCP à Cursor ?
Ouvrez ~/.cursor/mcp.json et ajoutez le serveur sous mcpServers, avec la configuration Cursor indiquée dans la section d'installation ci-dessus.
Que peut faire un agent IA avec le MCP server CronWatch ?
Un agent IA peut lister les tâches, lire les journaux d'erreurs, exécuter un test de santé et mettre les alertes en sourdine.
Différents langages de programmation peuvent-ils partager les mêmes données CronWatch ?
Oui, les packages de différents langages écrivent dans les mêmes tables, ce qui permet à une application Rails et un service Go de partager une seule base de données et un seul tableau de bord.