What is Model Context Protocol?

Model Context Protocol, or MCP, is a standard way for AI editors and assistants to connect to other services. Crontap uses it so you can manage your schedules, uptime monitors and heartbeats from the conversation.

Which editors and assistants work with Crontap MCP?

Setup is available for Cursor, Claude, Claude Code, ChatGPT, Codex CLI, VS Code desktop, Windsurf and Gemini CLI. Availability can still depend on your plan, region or workspace settings.

Yes. MCP access is included on Starter, Pro, Ultra and Custom plans. Starter requires no credit card and uses the same combined resource cap as the web app.

Is the MCP connection safe?

Crontap uses OAuth 2.1 with PKCE, one-hour access tokens, hashed token storage and explicit consent. You do not paste a raw API key into the client configuration.

Can I revoke an editor's access?

Yes. Connected MCP clients can be revoked from Crontap settings. Existing access for that connection then stops working.

Does MCP replace the Crontap API?

No. MCP is the conversational tool interface for supported AI clients. The REST API remains the direct programmatic interface and has its own plan availability and credentials.

Can MCP delete schedules, monitors or heartbeats?

Yes, when your editor supports changes and you approve the action. Deleting resources and rotating heartbeat URLs are marked as destructive so your editor can ask for confirmation.

What rate limits apply?

New schedules, monitors and heartbeats count toward your plan's existing limits. If the account is full, Crontap explains the available plan options.

Does Crontap MCP wake my agent on a schedule?

No. Your AI client starts each MCP conversation. Use your client's own automation for recurring prompts, Crontap schedules for recurring HTTP requests, and Crontap heartbeats to detect a job that stopped checking in.

Model Context Protocol

Run HTTP cron jobs. Monitor uptime and heartbeats in your AI editor.

Connect Crontap to Cursor, Claude, ChatGPT or another supported client. Then type what you need: schedule an endpoint, check a monitor or explain why a run failed.

Connect in 60 seconds Browse all tools

Server URLhttps://mcp.crontap.com/mcp

Cursor

Crontap

Approve Crontap

You choose what your editor can manage.

OAuth

Schedule an endpoint

Explain a failed run

Ready when you ask.

Supported editors

Cursor
Claude
ChatGPT
Codex CLI
VS Code
Windsurf
Gemini CLI

How it works

Connect once. Manage from the conversation.

Stay in your editor while Crontap handles the schedules and monitoring behind your requests.

Step 1

Add the remote server

Choose your editor below and paste in the Crontap server address.

https://mcp.crontap.com/mcp

Step 2

Approve access once

Step 3

Ask in plain English

Try “Run this endpoint every weekday morning” or “Why did last night’s job fail?” Review any important change before it happens.

Connect

Choose your editor and connect.

Follow the setup for the client you already use. You will sign in to Crontap in your browser, so there is no API key to paste into a config file.

Cursor

~/.cursor/mcp.json

{
  "mcpServers": {
    "crontap": {
      "url": "https://mcp.crontap.com/mcp"
    }
  }
}

Cursor IDE supports remote OAuth MCP servers and a one-click deeplink.

Open in Cursor Official client docs

Capabilities

What you can do from your editor.

Write the request as you normally would. Your editor works with Crontap, and you stay in control of changes such as deleting a schedule or rotating a heartbeat URL.

Send a weekday digest

You ask

“Every weekday at 8:00 AM Berlin time, POST to https://api.example.com/daily-digest.”

What happens

Crontap turns that into a cron schedule and shows the exact timezone and upcoming run times.

Keep watch over an endpoint

You ask

“Check https://example.com/health every five minutes, and help me investigate if it goes down.”

What happens

Crontap creates the uptime monitor and keeps the recent checks available when you need to trace an outage.

Catch a job that goes silent

You ask

“Set up a heartbeat for our nightly backup. It runs once a day, and 30 minutes late is a problem.”

What happens

Crontap gives the job a check-in URL and flags missed, late, failed and recovered runs.

Understand a failed sync

You ask

“Why did the last three billing-sync runs fail?”

What happens

Crontap reads the run history so you can see whether the request failed to send or the endpoint returned an error.

Check before adding more

You ask

“Do I have room for another uptime monitor on this plan?”

What happens

Crontap checks your current usage and limit, then explains the available options if the account is full.

Pause jobs during a deploy

You ask

“Pause the billing sync while we deploy, then resume it when I say the release is clear.”

What happens

Crontap pauses future runs without deleting the schedule or its history, then resumes the same configuration.

Tool reference

See what your editor can do.

Open a group to browse the available actions in plain language. Tool names stay visible when you need the exact reference, and an asterisk marks required information.

HTTP schedules

9 tools

list_schedules
See every scheduled HTTP job in the account.
Technical details
List the account's scheduled HTTP jobs and their current configuration and status. Use this before selecting an id for another schedule operation.
Information needed
No inputs
get_schedule
See one schedule's timing, request and current status.
Technical details
Read one scheduled HTTP job by its opaque id. Use list_schedules first when the id is unknown; the result includes configuration and current status.
Information needed
- scheduleId*
create_schedule
Schedule an HTTP request using cron or plain English.
Technical details
Create a scheduled HTTP request from either a cron expression or plain-English cadence. Text is previewed first; the result includes resolved cron, timezone, and next runs. If a plan cap blocks the request, use the returned four-option upgrade ladder to explain the smallest viable choices.
Information needed
- cron
- text
- url*
- timezone*
- method*
- headers
- body
- integrations
- label
- intervalLabel
- runOnce
- deleteAfterRun
- retryPolicy
update_schedule
Change a schedule without replacing its saved settings.
Technical details
Partially update a schedule's URL, cadence, timezone, HTTP request, or optional behavior. Omitted fields preserve the complete existing configuration, including hidden request secrets. If a plan cap blocks the request, use the returned four-option upgrade ladder to explain the smallest viable choices.
Information needed
- scheduleId*
- url
- cron
- timezone
- method
- headers
- body
- integrations
- label
- intervalLabel
- runOnce
- deleteAfterRun
- retryPolicy
delete_scheduleDestructive
Permanently remove a schedule and stop future runs.
Technical details
Permanently delete a scheduled HTTP job and stop all future runs. Confirm the target id with get_schedule before invoking this destructive operation.
Information needed
- scheduleId*
pause_schedule
Temporarily stop a schedule while keeping its history.
Technical details
Pause an existing schedule without deleting its configuration or history. Use resume_schedule later to allow future runs from the same schedule.
Information needed
- scheduleId*
resume_schedule
Start future runs for a paused schedule again.
Technical details
Resume a paused schedule so future HTTP runs can execute again. This preserves the existing cadence, request configuration, and retained history.
Information needed
- scheduleId*
preview_schedule
Check how a written cadence translates before saving.
Technical details
Convert a plain-English cadence into cron without saving anything. Returns the resolved timezone and next runs, or a clarification question that must be answered.
Information needed
- text*
- timezone*
get_schedule_history
Review recent runs and diagnose failures.
Technical details
Read cursor-paginated run history for one schedule, including outcomes needed to diagnose recent failures. Use the returned cursor to request another page.
Information needed
- scheduleId*
- limit*
- cursor

Tool	What it does	Information needed
`list_schedules`	See every scheduled HTTP job in the account. Technical details List the account's scheduled HTTP jobs and their current configuration and status. Use this before selecting an id for another schedule operation.	No inputs
`get_schedule`	See one schedule's timing, request and current status. Technical details Read one scheduled HTTP job by its opaque id. Use list_schedules first when the id is unknown; the result includes configuration and current status.	scheduleId*
`create_schedule`	Schedule an HTTP request using cron or plain English. Technical details Create a scheduled HTTP request from either a cron expression or plain-English cadence. Text is previewed first; the result includes resolved cron, timezone, and next runs. If a plan cap blocks the request, use the returned four-option upgrade ladder to explain the smallest viable choices.	cron text url* timezone* method* headers body integrations label intervalLabel runOnce deleteAfterRun retryPolicy
`update_schedule`	Change a schedule without replacing its saved settings. Technical details Partially update a schedule's URL, cadence, timezone, HTTP request, or optional behavior. Omitted fields preserve the complete existing configuration, including hidden request secrets. If a plan cap blocks the request, use the returned four-option upgrade ladder to explain the smallest viable choices.	scheduleId* url cron timezone method headers body integrations label intervalLabel runOnce deleteAfterRun retryPolicy
`delete_schedule` Destructive	Permanently remove a schedule and stop future runs. Technical details Permanently delete a scheduled HTTP job and stop all future runs. Confirm the target id with get_schedule before invoking this destructive operation.	scheduleId*
`pause_schedule`	Temporarily stop a schedule while keeping its history. Technical details Pause an existing schedule without deleting its configuration or history. Use resume_schedule later to allow future runs from the same schedule.	scheduleId*
`resume_schedule`	Start future runs for a paused schedule again. Technical details Resume a paused schedule so future HTTP runs can execute again. This preserves the existing cadence, request configuration, and retained history.	scheduleId*
`preview_schedule`	Check how a written cadence translates before saving. Technical details Convert a plain-English cadence into cron without saving anything. Returns the resolved timezone and next runs, or a clarification question that must be answered.	text* timezone*
`get_schedule_history`	Review recent runs and diagnose failures. Technical details Read cursor-paginated run history for one schedule, including outcomes needed to diagnose recent failures. Use the returned cursor to request another page.	scheduleId* limit* cursor

Uptime monitors

6 tools

list_monitors
See every uptime monitor and its current health.
Technical details
List uptime monitors with their endpoint, probe cadence, state, and recent reliability. Use this before selecting a monitor id for another operation.
Information needed
No inputs
get_monitor
See one monitor's endpoint, cadence and current state.
Technical details
Read one uptime monitor by its opaque id, including probe settings and current state. Use list_monitors first when the target id is unknown.
Information needed
- monitorId*
create_monitor
Start checking an HTTP endpoint for uptime.
Technical details
Create an uptime monitor that probes an HTTP endpoint on a fixed minute cadence and alerts on transitions. Pause later with update_monitor status paused. If a plan cap blocks the request, use the returned four-option upgrade ladder to explain the smallest viable choices.
Information needed
- name
- url*
- intervalMinutes*
- failureThreshold
- emailCooldownMinutes
- integrations
update_monitor
Change, pause or resume an uptime monitor.
Technical details
Partially update an uptime monitor's endpoint, cadence, thresholds, notifications, or status. Set status paused to pause and pending to resume it. If a plan cap blocks the request, use the returned four-option upgrade ladder to explain the smallest viable choices.
Information needed
- monitorId*
- name
- url
- intervalMinutes
- failureThreshold
- emailCooldownMinutes
- integrations
- status
delete_monitorDestructive
Permanently remove a monitor and stop its probes.
Technical details
Permanently delete an uptime monitor and stop all future probes. Confirm the target with get_monitor before invoking this destructive operation.
Information needed
- monitorId*
get_monitor_probes
Review a monitor's checks for a specific day.
Technical details
Read retained probe results for one uptime monitor on a required UTC day. Use this to diagnose outages, response codes, and recovery timing.
Information needed
- monitorId*
- day*

Tool	What it does	Information needed
`list_monitors`	See every uptime monitor and its current health. Technical details List uptime monitors with their endpoint, probe cadence, state, and recent reliability. Use this before selecting a monitor id for another operation.	No inputs
`get_monitor`	See one monitor's endpoint, cadence and current state. Technical details Read one uptime monitor by its opaque id, including probe settings and current state. Use list_monitors first when the target id is unknown.	monitorId*
`create_monitor`	Start checking an HTTP endpoint for uptime. Technical details Create an uptime monitor that probes an HTTP endpoint on a fixed minute cadence and alerts on transitions. Pause later with update_monitor status paused. If a plan cap blocks the request, use the returned four-option upgrade ladder to explain the smallest viable choices.	name url* intervalMinutes* failureThreshold emailCooldownMinutes integrations
`update_monitor`	Change, pause or resume an uptime monitor. Technical details Partially update an uptime monitor's endpoint, cadence, thresholds, notifications, or status. Set status paused to pause and pending to resume it. If a plan cap blocks the request, use the returned four-option upgrade ladder to explain the smallest viable choices.	monitorId* name url intervalMinutes failureThreshold emailCooldownMinutes integrations status
`delete_monitor` Destructive	Permanently remove a monitor and stop its probes. Technical details Permanently delete an uptime monitor and stop all future probes. Confirm the target with get_monitor before invoking this destructive operation.	monitorId*
`get_monitor_probes`	Review a monitor's checks for a specific day. Technical details Read retained probe results for one uptime monitor on a required UTC day. Use this to diagnose outages, response codes, and recovery timing.	monitorId* day*

Heartbeats

7 tools

list_heartbeats
See heartbeat checks and their recent activity.
Technical details
List heartbeat checks and daily activity without exposing their secret ping URLs. Use this before selecting a heartbeat id for another operation.
Information needed
- window*
get_heartbeat
See one heartbeat's timing and current state.
Technical details
Read one heartbeat's configuration and current state without exposing its secret ping URL. Rotate the token when a new URL must be obtained.
Information needed
- heartbeatId*
create_heartbeat
Create a check-in URL for an existing job.
Technical details
Create a dead-man's-switch heartbeat for an existing job and return its secret ping URL once with a curl command. Crontap monitors pings; it does not wake agents. If a plan cap blocks the request, use the returned four-option upgrade ladder to explain the smallest viable choices.
Information needed
- name
- periodMinutes*
- graceMinutes
- emailCooldownMinutes
- webhookUrl
- integrations
update_heartbeat
Change, pause or resume a heartbeat check.
Technical details
Partially update heartbeat timing, notifications, or status. Set status paused to suspend deadlines and pending to resume from a clean state. If a plan cap blocks the request, use the returned four-option upgrade ladder to explain the smallest viable choices.
Information needed
- heartbeatId*
- name
- periodMinutes
- graceMinutes
- emailCooldownMinutes
- webhookUrl
- integrations
- status
delete_heartbeatDestructive
Permanently remove a heartbeat and its check-in URL.
Technical details
Permanently delete a heartbeat and invalidate its ping endpoint. Confirm the target with get_heartbeat before invoking this destructive operation.
Information needed
- heartbeatId*
rotate_heartbeat_tokenDestructive
Replace a heartbeat's secret check-in URL.
Technical details
Generate a new secret heartbeat ping URL and immediately invalidate the old URL and its fail variant. Update the monitored job before its next deadline.
Information needed
- heartbeatId*
get_heartbeat_pings
Review successful and failed check-ins for a day.
Technical details
Read retained success, failure, and test ping events for one heartbeat on a required UTC day. Use this to diagnose missed or failed check-ins.
Information needed
- heartbeatId*
- day*

Tool	What it does	Information needed
`list_heartbeats`	See heartbeat checks and their recent activity. Technical details List heartbeat checks and daily activity without exposing their secret ping URLs. Use this before selecting a heartbeat id for another operation.	window*
`get_heartbeat`	See one heartbeat's timing and current state. Technical details Read one heartbeat's configuration and current state without exposing its secret ping URL. Rotate the token when a new URL must be obtained.	heartbeatId*
`create_heartbeat`	Create a check-in URL for an existing job. Technical details Create a dead-man's-switch heartbeat for an existing job and return its secret ping URL once with a curl command. Crontap monitors pings; it does not wake agents. If a plan cap blocks the request, use the returned four-option upgrade ladder to explain the smallest viable choices.	name periodMinutes* graceMinutes emailCooldownMinutes webhookUrl integrations
`update_heartbeat`	Change, pause or resume a heartbeat check. Technical details Partially update heartbeat timing, notifications, or status. Set status paused to suspend deadlines and pending to resume from a clean state. If a plan cap blocks the request, use the returned four-option upgrade ladder to explain the smallest viable choices.	heartbeatId* name periodMinutes graceMinutes emailCooldownMinutes webhookUrl integrations status
`delete_heartbeat` Destructive	Permanently remove a heartbeat and its check-in URL. Technical details Permanently delete a heartbeat and invalidate its ping endpoint. Confirm the target with get_heartbeat before invoking this destructive operation.	heartbeatId*
`rotate_heartbeat_token` Destructive	Replace a heartbeat's secret check-in URL. Technical details Generate a new secret heartbeat ping URL and immediately invalidate the old URL and its fail variant. Update the monitored job before its next deadline.	heartbeatId*
`get_heartbeat_pings`	Review successful and failed check-ins for a day. Technical details Read retained success, failure, and test ping events for one heartbeat on a required UTC day. Use this to diagnose missed or failed check-ins.	heartbeatId* day*

Account and discovery

2 tools

get_account_usage
Check the current plan, usage and available capacity.
Technical details
Read the current Crontap tier, resource usage, caps, feature access, and public plan options. Use it before creating resources when capacity is uncertain.
Information needed
No inputs
list_timezones
Find an accepted timezone name for a schedule.
Technical details
Find exact IANA timezone names accepted by schedule creation and preview using an optional case-insensitive query. Without one, returns a bounded categorized sample.
Information needed
- query

Tool	What it does	Information needed
`get_account_usage`	Check the current plan, usage and available capacity. Technical details Read the current Crontap tier, resource usage, caps, feature access, and public plan options. Use it before creating resources when capacity is uncertain.	No inputs
`list_timezones`	Find an accepted timezone name for a schedule. Technical details Find exact IANA timezone names accepted by schedule creation and preview using an optional case-insensitive query. Without one, returns a bounded categorized sample.	query

Security model

Approve access without sharing an API key.

Supported editors and assistants open a familiar authorization flow, with short-lived credentials and account-level revocation.

OAuth 2.1 with PKCE: Your client opens Crontap consent. Passwords and raw API keys never go into the MCP configuration.
Short-lived access: Access tokens expire after one hour. Stored token material is hashed, not kept as reusable plaintext.
Separate credentials: MCP tokens are scoped to the MCP server and cannot be reused as raw Crontap API keys.
Visible control: Delete operations are marked destructive, and connections can be revoked from Crontap settings.

Your editor

Crontap

Approve CrontapReview access before connecting.

OAuth 2.1 + PKCE

View schedules and recent run history
Manage uptime monitors and heartbeats
Revoke this connection from Crontap settings

Access expires in 1 hour

Pricing

MCP is included on every plan.

Start free. Your existing combined resource cap applies across schedules, uptime monitors and heartbeats.

Starter

Free

$0 / month

2 combined items, with MCP included and no card required.

Pro

Starts at

$2.99 / month

Minute cadence and larger combined limits.

Ultra

Starts at

$11.99 / month

25+ combined items, public API access and priority support.

Custom

Tailored

Talk to us

1,000+ items, SSO, dedicated support and a contractual SLA.

Compare every plan →

Editor guides are coming.

We are preparing step-by-step setup guides for each supported editor. These useful scheduling and monitoring guides are available now.

Useful guides available now

FAQ

Crontap MCP questions, answered.

What is Model Context Protocol?: Model Context Protocol, or MCP, is a standard way for AI editors and assistants to connect to other services. Crontap uses it so you can manage your schedules, uptime monitors and heartbeats from the conversation.
Which editors and assistants work with Crontap MCP?: Setup is available for Cursor, Claude, Claude Code, ChatGPT, Codex CLI, VS Code desktop, Windsurf and Gemini CLI. Availability can still depend on your plan, region or workspace settings.
Is Crontap MCP free?: Yes. MCP access is included on Starter, Pro, Ultra and Custom plans. Starter requires no credit card and uses the same combined resource cap as the web app.
Is the MCP connection safe?: Crontap uses OAuth 2.1 with PKCE, one-hour access tokens, hashed token storage and explicit consent. You do not paste a raw API key into the client configuration.
Can I revoke an editor's access?: Yes. Connected MCP clients can be revoked from Crontap settings. Existing access for that connection then stops working.
Does MCP replace the Crontap API?: No. MCP is the conversational tool interface for supported AI clients. The REST API remains the direct programmatic interface and has its own plan availability and credentials.
Can MCP delete schedules, monitors or heartbeats?: Yes, when your editor supports changes and you approve the action. Deleting resources and rotating heartbeat URLs are marked as destructive so your editor can ask for confirmation.
What rate limits apply?: New schedules, monitors and heartbeats count toward your plan's existing limits. If the account is full, Crontap explains the available plan options.
Does Crontap MCP wake my agent on a schedule?: No. Your AI client starts each MCP conversation. Use your client's own automation for recurring prompts, Crontap schedules for recurring HTTP requests, and Crontap heartbeats to detect a job that stopped checking in.

Ready to fix it?

Point Crontap at any URL. Pick any cron. Done.

WordPress, Shopify, Railway, Cloud Run, Vercel, HubSpot, Ghost, your own box. If it answers HTTP, Crontap can drive it on a clock you can read, in the timezone that actually matters, and page you when something breaks.

Get started Connect your editor

Free forever tier ・ No credit card required

GET

/wp-cron.php?doing_wp_cron=1

Running

Your next schedule

Schedule

"every 5 minutes"

in 23s