Ntfy
Send push notifications to your phone or desktop via PUT/POST
Verdict
Common use cases
- Alert on-call engineers when deployments fail
- Notify team channels about CI pipeline status
- Send reminders for scheduled maintenance windows
- Route customer support escalations to mobile devices
- Broadcast system health checks to Slack alternatives
Integration
- Vendor
- Ntfy
- Category
- other
- Auth
- API_KEY
- Tools
- 22
- Composio slug
ntfy
Tools
- Check NTFY Service Health
Tool to check the health status of the ntfy service. Use when you need to verify if the ntfy service is operational and responding correctly.
- Create NTFY Account
Tool to register a new user account on ntfy. Use when you need to create a new user account on the ntfy.sh hosted service. Note: This endpoint is not available on self-hosted instances unless signup is explicitly enabled in the server confi
- Create Web Push Subscription
Tool to register a web push subscription for browser notifications. Use when you need to enable push notifications from ntfy topics through a browser's native push notification system.
- Delete Web Push Subscriptiondestructive
Tool to unregister a web push subscription from the ntfy server. Use when you need to remove browser push notifications for a previously registered endpoint.
- Fetch Cached Messages
Tool to fetch cached messages from a ntfy topic. Use when you need to retrieve previously sent messages stored on the server. Supports filtering by time (duration or timestamp), message ID, content, title, priority, and tags. Set poll=1 to
- Fetch Latest Message from Topic
Tool to fetch the most recent message from a topic's cache. Use when you need to retrieve the latest message without subscribing to the topic stream.
- Fetch Scheduled Messages
Tool to fetch messages scheduled for later delivery from a topic. Use when you need to retrieve messages that are set to be delivered at a future date. The poll=1 and scheduled=1 parameters are automatically set to retrieve cached scheduled
- Get Account Information
Tool to retrieve account data for authenticated user or anonymous user. Use when you need to get account information, limits, statistics, or tier details.
- Get File Attachment Metadata
Tool to get file attachment metadata from a message without downloading the file content. Use when you need to check file size, content type, or verify file existence in the ntfy attachment cache.
- Get Server Statistics
Tool to retrieve server statistics including message counts and publishing rates. Use when you need to monitor ntfy server activity and get metrics about message volume.
- Get Service Tiers
Tool to list all available ntfy service tiers with their limits and features. Use when you need to retrieve subscription tier information including pricing and usage limits. Requires that payments are enabled on the server.
- Poll Messages from Topic
Tool to poll for messages from an ntfy topic without maintaining a long-standing connection. Use when you need to retrieve cached messages and have the connection close immediately after delivery. The connection ends after reading all avail
- Publish Message as JSON to NTFY
Tool to publish messages as JSON to ntfy. Use when you need to send notifications with all parameters in the request body, especially useful for integrations that cannot add custom headers.
- Publish Message to Topic
Tool to publish a message to a ntfy topic. Use when you need to send notifications or alerts to a topic. Topics are created dynamically if they don't exist.
- Publish Message to Topic (PUT)
Tool to publish a message to a topic using PUT method. Use when you need to send notifications to subscribers of a topic. Supports various options like priority, tags, attachments, and scheduled delivery.
- Publish Message via GET
Tool to publish messages to ntfy via GET request with URL parameters. Use when PUT/POST methods are unavailable or for simple webhook integration. Supports all message parameters as query strings. Without parameters, sends 'triggered' as me
- Send Message via Webhook
Tool to send messages via webhook endpoint using simple GET request. Use when you need a simple webhook-style integration or for clients with limited HTTP support.
- Subscribe to Multiple NTFY Topics
Tool to subscribe to multiple ntfy topics simultaneously using comma-separated topic list. Use when you need to receive messages from multiple topics in a single API call. Returns cached messages when used with poll=true parameter.
- Subscribe to NTFY Topic with Filters
Tool to subscribe to a ntfy topic with filters based on message fields (id, message, title, priority, tags). Use when you need to retrieve specific messages from a topic using filter criteria. Returns matching messages from the topic as a J
- Subscribe to Topic (JSON Stream)
Tool to subscribe to a ntfy topic and receive messages as JSON stream. Use when you need to retrieve cached messages from a topic. The action polls the topic and returns immediately with cached messages.
- Subscribe to Topic (Raw Stream)
Tool to subscribe to a topic and receive message bodies as raw text stream. Use when you need to retrieve messages without metadata like priority, tags, or titles. Each line in the response contains only the message body. Empty lines repres
- Trigger NTFY Webhook
Tool to trigger a webhook to publish a message to an ntfy topic via simple HTTP GET request. Use when you need to send notifications through webhooks without requiring a POST body. Sends a default 'triggered' message if no custom message is
Setup
Setup guide
- 11. In Switchy, open your workspace settings and navigate to Integrations > Add MCP. 2. Select Ntfy from the list and choose API Key as the auth method. 3. If you're using the hosted ntfy.sh service, create an account at ntfy.sh and generate an API key from your account settings; if you're self-hosting, retrieve your key from your instance's admin panel. 4. Paste the API key into Switchy and specify your ntfy server URL (https://ntfy.sh for hosted, or your self-hosted domain). 5. Click Connect and wait for Switchy to verify the connection by calling the health check endpoint. 6. Open any Space, type '@Ntfy check service health' and send — if you see a status response, the integration is live. 7. To publish your first notification, try '@Ntfy publish a test message to topic my-team-alerts' (replace 'my-team-alerts' with a topic name you control).
What teammates see: by default, memories from Ntfy are scoped to the Space (PROJECT visibility) - you can mark any memory PRIVATE or share it ORG-wide.
Works well with
Top models
Compatibility data appears once enough Spaces have used this MCP together with a given model.
How Switchy teams use it
Starter prompts
Check Service Health
@Ntfy check the health status of the ntfy service and confirm it's responding correctly.Open in a Space →
Publish Deployment Alert
@Ntfy publish a message to the topic 'prod-deploys' with the title 'Deployment Complete' and the body 'API v2.3.1 is live in us-east-1'.Open in a Space →
Fetch Recent Notifications
@Ntfy fetch the last 10 cached messages from the topic 'security-alerts' and summarize any critical issues.Open in a Space →
List Scheduled Messages
@Ntfy fetch all scheduled messages from the topic 'maintenance-reminders' and list their delivery times.Open in a Space →
Get Account Limits
@Ntfy retrieve my account information including message limits, current usage, and tier details.Open in a Space →
Example outputs
Illustrative - representative of the model's voice and quality, not literal recordings.
@Ntfy check if the service is healthy and then fetch the last 5 cached messages from the 'deployment-alerts' topic
Service health check passed — ntfy is operational. Retrieved 5 cached messages from 'deployment-alerts': 1. [2 hours ago] 'Production deploy started: v2.4.1' 2. [1 hour ago] 'Database migration complete' 3. [45 min ago] 'Health checks passing on all nodes' 4. [30 min ago] 'Deploy completed successfully' 5. [15 min ago] 'Rollback triggered: API latency spike' All messages include timestamps and priority levels. The most recent message indicates a rollback event that may require attention.
This example shows ntfy's core read capability: verifying service availability and retrieving message history. Cached messages are stored server-side for a limited time (typically 12-48 hours depending on tier). You'll need read access to the topic, and message retention depends on your ntfy instance's configuration.
@Ntfy send a high-priority notification to the 'oncall-team' topic with the title 'Database CPU at 95%' and message 'RDS instance db-prod-01 needs immediate attention'
Notification sent successfully to 'oncall-team' topic: • Title: 'Database CPU at 95%' • Message: 'RDS instance db-prod-01 needs immediate attention' • Priority: High (4/5) • Delivery: Immediate • Message ID: msg_8x2k9p4n All subscribers to 'oncall-team' will receive this alert on their registered devices. The high priority ensures prominent notification display and sound on mobile devices.
This demonstrates ntfy's primary use case: real-time alerting. The notification is delivered immediately to all topic subscribers. Be aware that ntfy topics are public by default unless you configure access control with authentication. High-priority messages may bypass Do Not Disturb settings on some devices, so use this level judiciously.
@Ntfy fetch all scheduled messages from 'maintenance-reminders' and summarize which systems have upcoming maintenance windows this week
Retrieved 4 scheduled messages from 'maintenance-reminders'. Here's this week's maintenance schedule: **Tuesday 3 AM UTC**: Payment gateway upgrade (2-hour window) **Thursday 1 AM UTC**: CDN certificate renewal (30-minute window) **Saturday 11 PM UTC**: Database index rebuild on analytics cluster (4-hour window) **Sunday 2 AM UTC**: Load balancer firmware update (1-hour window) All maintenance windows are scheduled during low-traffic periods. The database rebuild on Saturday is the longest disruption and may impact reporting dashboards.
This example pairs ntfy's scheduled message retrieval with AI synthesis to create an actionable summary. Scheduled messages are useful for planning ahead, but they're stored separately from the regular message cache. The AI can help you spot conflicts or patterns across multiple scheduled notifications that might not be obvious from raw message data.
Use-case deep-dives
When Ntfy beats Slack for CI/CD notifications
A 3-person dev team runs nightly builds that fail 20% of the time. They tried Slack webhooks but got alert fatigue from channel noise and missed critical failures in the morning. Ntfy wins here because it routes build status to personal devices without polluting shared channels. The team uses the Fetch Cached Messages tool to review overnight failures during standup, and Check NTFY Service Health confirms the notification pipeline is live before deployments. The trade-off: if your team needs threaded discussion on failures or wants alerts tied to Linear tickets, Ntfy stops short—it's a one-way push system. But for teams under 10 people who just need reliable 'something broke' pings without Slack's $8/user/month cost, Ntfy delivers the alert and gets out of the way.
Ntfy for on-call routing when PagerDuty feels heavy
A 5-person support team handles 40 tickets a day across Zendesk and email. High-priority tickets (refund requests, data loss reports) need immediate attention, but PagerDuty's $19/user/month pricing doesn't fit their budget. They use Ntfy to push escalations to the on-call rep's phone via Create Web Push Subscription, with the Fetch Latest Message tool confirming delivery during handoffs. The system works because their escalation volume is low (3-5 per day) and they don't need acknowledgment workflows or incident timelines. The boundary: once you're triaging more than 10 urgent tickets daily or need audit logs for compliance, Ntfy's simplicity becomes a liability. For lean support teams with straightforward escalation rules, Ntfy covers the 'get this person's attention now' job for free.
When Ntfy handles editorial calendar nudges better than email
A 4-person content team publishes 12 blog posts a month with staggered deadlines. They tried Google Calendar reminders but missed notifications buried in email, and Asana's notification settings were too coarse. Ntfy's Fetch Scheduled Messages tool lets them queue deadline reminders 48 hours and 2 hours before each post is due, delivered as phone notifications that bypass inbox clutter. The Create NTFY Account tool provisions each writer with a personal topic for their assignments. This works because their cadence is predictable and they don't need two-way confirmation that someone saw the reminder. The limit: if your editorial workflow requires proof of acknowledgment or integrates with CMS approval gates, Ntfy won't close the loop. For small content teams who just need reliable 'your draft is due' pings without enterprise notification tooling, Ntfy keeps deadlines visible for zero cost.
Frequently asked
What does the Ntfy MCP do in Switchy?
The Ntfy MCP lets your AI agents send and receive push notifications through Ntfy's pub-sub service. Agents can publish messages to topics, fetch cached or scheduled messages, manage subscriptions, and check service health. It's useful for alerting your team when an agent completes a task, encounters an error, or needs human input without polling Slack or email.
Do I need a paid Ntfy account to use this MCP?
No. The MCP works with Ntfy's free tier at ntfy.sh or a self-hosted instance. You'll need an API key if your instance requires authentication. The MCP can create accounts and manage web push subscriptions, but most teams just generate a token from an existing account and paste it into Switchy's auth field.
Can the MCP send attachments or images in notifications?
Yes, if the underlying Ntfy instance supports it. The MCP exposes Ntfy's full message API, which includes attachment URLs, inline images, action buttons, and priority levels. Your agent can reference a file URL in the message payload. Switchy doesn't host the files—you'll need to upload them elsewhere first or use Ntfy's attachment upload endpoint.
Why use this instead of just calling Ntfy's HTTP API directly?
The MCP wraps Ntfy's REST endpoints so your agents can discover and call them without you writing fetch code or managing tokens in prompts. It also handles polling, scheduled message retrieval, and subscription lifecycle in a single tool set. If you're already piping agent output to a custom script, the API is simpler.
Who on the team should connect the Ntfy MCP?
Whoever runs your Ntfy instance or holds the API key. If you're using ntfy.sh, any team member can create a free account and share the token in Switchy. The MCP doesn't consume Switchy seats or count against message limits—it just forwards notifications to whatever topics you configure in your agent prompts.