claude-cron
MCP server that allows AI coding assistants to schedule deferred tasks such as reminders, shell commands, and AI prompts, executing them as an OS daemon.
README
Claude Cron
Deferred tasks for AI coding assistants. MCP server + daemon that allows any AI coding assistant to schedule tasks for the future.
How does this compare to
/loop? Claude Code 2.1.71 added/loopfor in-session recurring tasks. It's great for ephemeral monitoring but dies when you close the terminal. claude-cron runs as an OS daemon — your tasks survive reboots and execute while you sleep. Read the full comparison.
Quickstart
bun install
bun test # 38 tests
bun run typecheck # tsc --noEmit
Claude Code Configuration
Add to ~/.claude/settings.json:
{
"mcpServers": {
"claude-cron": {
"command": "bun",
"args": ["run", "/Users/fernando/code/claude-cron/src/index.ts"],
"env": {
"CLAUDE_CRON_DB": "/Users/fernando/.claude/claude-cron.sqlite",
"GMAIL_ADDRESS": "frr@keepcoding.io",
"GMAIL_APP_PASSWORD": "op://FRR DEV/...",
"NTFY_TOPIC": "claude-cron-frr-xxx",
"NTFY_URL": "https://ntfy.sh"
}
}
}
}
Architecture
AI Assistant (Claude Code / Codex / Copilot)
│ stdio (MCP protocol)
▼
MCP Server (src/index.ts) ← Task CRUD
│ SQLite (WAL mode)
▼
Runner / Daemon (src/runner.ts) ← Loop: executes due tasks
├─ reminder → notification + email + push
├─ shell → Bun.spawn(command)
├─ claude → claude -p --output-format text "prompt"
├─ codex → codex --quiet "prompt"
└─ copilot → gh copilot suggest "prompt"
MCP Tools
| Tool | Description |
|---|---|
schedule_task |
Schedule a task for the future |
list_tasks |
List tasks (filter by status, upcoming, tags) |
cancel_task |
Cancel a task by ID |
run_now |
Mark a task for immediate execution |
install_daemon |
Register the runner as an OS service |
schedule_task
{
name: "Nightly tests",
type: "shell", // reminder | shell | claude | codex | copilot
when: "daily at 3:00", // see "When expressions" below
payload: {
command: "cd ~/code/bfclaude && make test", // shell
// message: "...", // reminder
// prompt: "...", // claude/codex/copilot
// model: "haiku",
// workingDir: "/path",
// maxBudgetUsd: 0.10,
},
notify: ["notification", "email", "push"],
notifyTo: "frr@keepcoding.io",
tags: ["ci"],
}
when Expressions
| Type | Examples |
|---|---|
| Natural language | "tomorrow 9:00", "in 30 minutes", "next friday at 3pm" |
| Cron (5 fields) | "0 9 * * 1", "*/5 * * * *" |
| ISO 8601 | "2026-03-01T09:00:00Z", "2026-03-01" |
| Shortcuts | "every monday at 8", "daily at 9:00", "every 5 minutes" |
Parsing: chrono-node (natural) + croner (cron) + regex (shortcuts). Shortcuts are evaluated before chrono-node so that "every monday at 8" is not interpreted as a single date.
Structure
src/
├── index.ts # MCP server entry point (5 tools)
├── runner.ts # Daemon: oneshot (launchd) or --loop (dev)
├── queue.ts # SQLite: schema, CRUD, getDueTasks
├── scheduler.ts # parseWhen(): chrono-node + croner + shortcuts
├── adapters/
│ ├── types.ts # AIAdapter interface
│ ├── claude.ts # claude -p --output-format text
│ ├── codex.ts # codex --quiet (stub)
│ └── copilot.ts # gh copilot suggest (stub)
├── notify/
│ ├── types.ts # NotificationAdapter interface
│ ├── desktop.ts # node-notifier (cross-platform)
│ ├── email.ts # Direct SMTP to Gmail (TLS 465)
│ └── push.ts # ntfy.sh (mobile + Apple Watch)
└── daemon/
├── types.ts # DaemonInstaller interface
├── detect.ts # process.platform → installer
├── launchd.ts # macOS: ~/Library/LaunchAgents/
├── systemd.ts # Linux: ~/.config/systemd/user/
└── schtasks.ts # Windows: Task Scheduler
Runner
Two modes:
# Oneshot (for launchd StartInterval / systemd timer)
bun run src/runner.ts
# Continuous loop (for dev or KeepAlive)
bun run src/runner.ts --loop
The daemon reads SQLite every 60s (configurable via CLAUDE_CRON_INTERVAL), executes due tasks, and sends notifications.
Cron tasks are automatically rescheduled after completion (next_run recalculated). One-off tasks transition to completed or failed.
Daemon
# Install (from Claude Code via MCP tool install_daemon, or manually):
# macOS:
bun run src/daemon/launchd.ts # generates plist + launchctl load
# Verify
launchctl list | grep claude-cron
tail -f ~/Library/Logs/claude-cron.log
# Uninstall
launchctl unload ~/Library/LaunchAgents/com.claude-cron.runner.plist
Notifications
| Channel | Requirements | Env vars |
|---|---|---|
| Desktop | node-notifier (included) | — |
| Gmail App Password | GMAIL_ADDRESS, GMAIL_APP_PASSWORD |
|
| Push (ntfy) | ntfy app on mobile/watch | NTFY_TOPIC, NTFY_URL (opt), NTFY_TOKEN (opt) |
SQLite Schema
tasks (
id TEXT PRIMARY KEY, -- nanoid 12 chars
name, type, status, -- type: reminder|shell|claude|codex|copilot
run_at INTEGER, -- Unix ts (one-off)
cron TEXT, -- Cron expression (recurring)
payload TEXT, -- JSON per type
notify_via TEXT, -- JSON array: ["notification","email","push"]
notify_to TEXT, -- Destination email
next_run, last_run, -- Timestamps
run_count, last_result, -- Executions + JSON result
created_at, created_by, tags -- Metadata
)
Dependencies
@modelcontextprotocol/sdk— MCP protocolchrono-node— Natural language → Datecroner— Cron parsing + next runnanoid— Short IDsnode-notifier— Cross-platform desktop notificationszod— MCP input validationbun:sqlite— built-in, zero deps
TODO
- [ ] Configure MCP in real settings.json and test end-to-end
- [ ] Install daemon with
install_daemonand verify full cycle - [ ] Set up ntfy for push to mobile/watch
- [ ] npm publish (name TBD)
- [ ] CI with tests on macOS + Linux
- [ ] README with real usage GIFs
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.