Starter prompt
This is Status, a public status page with uptime bars and incident updates that checks my sites every hour. You, my agent, can write incident updates for me. Step 1. If this is the Template, copy it for me with copy_app and work on my copy. Read it with get_app_source. It shows a sample company until the first change. getAdmin and getAgentInbox are members-only, so call them with mutate_app. Step 2. Ask me one at a time: my company name, the parts of my product customers care about (like Website, Dashboard, API), and for each one a public https URL that returns 200 when it is healthy, if it has one. Step 3. Set it up. Call startFresh with confirm true, my company name, and components (name, monitorUrl). For each monitor URL, add its exact origin (https://host, no path) to manifest.permissions.server.fetch with update_app, keeping the existing entries. Make the app Public with set_public. Step 4. Create an hourly Routine on checkMonitors, call checkMonitors once now, and show me the results from getAdmin. Step 5. When a check fails twice, the app opens an incident and leaves a ping for you in getAgentInbox. Check it when I ask about status. Write updates with agentWrite (actions update and resolve) in plain, short sentences for customers, then call ackPing. Optional, for instant pings: save your webhook URL with saveSettings, add its origin to the allowed list, tell me any secret name to set in App settings, Secrets (never ask me to paste it here), call sendTestPing, and create an hourly Routine on retryPings. Step 6. Optional: create a Webhook on agentWrite so my CI or another monitor can open and resolve incidents. Finish with the public link and how to post an incident from the app's admin bar.
What it does
A status page your customers can bookmark, in the style of Statuspage. It shows each part of your product with 90 days of uptime, the incidents you are working on, past incidents with every update, and scheduled maintenance.
- A green "All Systems Operational" banner, or the worst current state when something is wrong
- Incidents with Investigating, Identified, Monitoring, and Resolved updates
- Scheduled maintenance with a time window
- Hourly checks on the sites you choose. When a check fails twice, it marks the part down and opens an incident for you.
How to use it
- Connect your agent to Charming. Claude, ChatGPT, and other agents that support MCP can do it; the setup steps are at charm.ing/docs.
- Paste the starter prompt from this page into your agent. It copies the app to your account and starts the setup.
- Answer your agent's questions: your company name, the parts of your product to show (like Website, Dashboard, API), and a public https address for each one that returns 200 when it is healthy.
- Your agent sets up the page, allows each address for checks, makes the page Public, and runs the first check. Share the link with your customers.
- When something breaks, say "post an incident: the API is slow" and your agent writes it. Say "we found the cause" or "it is fixed" and it adds the update or resolves the incident. You can also post from the admin bar at the top of the page.
- When a check fails twice, the app opens an incident and leaves a ping in your agent's inbox. Ask your agent "any status pings?" and it writes the customer update for you.
Your agent writes the update
When a check fails, your agent gets the details and can post the first update while you are still opening your laptop. It posts later updates and resolves the incident the same way. A CI job or another monitor can open and resolve incidents through a Webhook.
What you need
Nothing for the page itself: make the app Public and share the link. Checks need your agent to add each site to the app's allowed list. Pings to your agent are optional, and your agent can also read the inbox whenever you ask.
What comes next
- Add a Webhook so your CI or another monitor can open and resolve incidents.
- Add an hourly Routine so your agent checks the inbox and writes updates without being asked.
- Email and text subscriptions are not built in yet. The Subscribe button shows your page link and an optional "Follow us" link, which you set in Manage.
More from Tambo
A to-do list you and your AI agent both work from. It replaces the markdown file, sync script, and…
To-Do Board
A to-do list you and your AI agent both work from. It replaces the markdown file, sync script, and chat relay you would wire up so an agent can see and tick your list. For anyone who runs their week through an agent.
FreshBooks-style invoicing for freelancers: build an invoice in a minute, send your client a link,…
Invoices
FreshBooks-style invoicing for freelancers: build an invoice in a minute, send your client a link, see when they open it, and let your agent chase late ones.
A public changelog for small product teams that ship every week, in place of a paid changelog tool.…
Changelog
A public changelog for small product teams that ship every week, in place of a paid changelog tool. Merged pull requests arrive as drafts, your agent writes the release notes, and you publish.