Developers app
AI agent status on your desk
The AI agent status app shows what Claude Code or any coding agent is doing on Perchi Matrix: idle, working, needs input, done or error. The agent's own hooks post the state to a webhook URL, so nothing else from the session is read. The display flashes when the agent needs you and celebrates when it finishes, within 1 second.
- Available
- Webhook
- Works on Perchi Matrix
- Works on Perchi
App id agent.status. Checked against Claude Code and other agents on .
What it shows
What does the AI agent status app show?
Shows when Claude Code or any coding agent is working, needs your input or is done, from the agent's own hooks, with a flash when it needs you.
The status layout shows the agent's state beside an animated robot icon: idle, working, needs input, done or error, scrolling when the word does not fit. A second layout shows today's token count with a TOK label, reset daily. The state changes the instant the hook fires, because the agent posts it to a webhook URL rather than anything being polled.
For Claude Code the dashboard shows a ready-made hooks recipe: the Notification hook sends {"state":"needs_input"}, the Stop hook sends {"state":"done"}, and a UserPromptSubmit hook sends working. Any agent, script or CI job that can make an HTTP POST can do the same; the app understands aliases such as busy, thinking, waiting and finished, so you rarely have to match the exact words.
Only the state and an optional token count are read from the request; nothing about your prompts, files or code is sent, because the hook itself decides what to post. The needs_input state plays a flash with INPUT so a waiting agent is noticed from across the room; done plays confetti.
How it connects
What do you need to connect AI agent status?
Paste a webhook URL. Paste the webhook URL and token into your agent's hooks. The Claude Code recipe is ready to copy.
A webhook only ever flows one way, from the provider to MyPerchi: the provider signs each delivery with the secret you set, duplicates are answered and ignored, and the delivery log keeps 7 days of history. The developers page covers presets, filters and limits.
Speed
How fast does AI agent status update?
Webhook only: the state changes within a second of the hook's POST, over the signed WebSocket every display keeps open. Nothing is polled, and a display that has not heard from the agent for a while marks the state stale rather than showing an old working.
Style
How does AI agent status look on the matrix?
The robot icon, animated, beside the state word is the default, scrolling when needs input does not fit in 15 pixels; the TOK layout shows today's tokens. needs_input plays a flash with INPUT and done plays confetti with DONE; both texts and effects are editable, and a threshold can turn error red.
The preview at the top of this page is the default AI agent status scene on a 24x8 Perchi Matrix, reading WAIT. The apps page explains how freshness badges work, and the Perchi Matrix page lists every font, colour and effect in the style panel.
On Perchi
What does Perchi do with AI agent status?
needs input is a curious moment on Perchi and done a proud one, each with the state on the card. Perchi has no speaker, so the signal is the face, which is the point.
A moment is Perchi's reaction to a real event: the face changes mood first, then a small card with the value slides in beside it and the face returns. Moments queue three deep, never interrupt a Phone Nap stage change, and wait silently during a block when do-not-disturb is on (the default). See which real events Perchi celebrates and how Phone Nap works.
Setup
Set up AI agent status in three steps
Everything happens in the web dashboard from any browser; there is no app to install on your phone and no account wall before the first pixel.
- STEP 1
Install and create the automation
Install AI agent status, then add an automation with a token header targeting it, with Set a value and no path.
- STEP 2
Copy the Claude Code recipe
The automation page shows the hooks JSON with your URL and token filled in; paste it into your Claude Code settings.
- STEP 3
Run a task
Submit a prompt and the display reads working; when Claude Code asks a question it flashes INPUT; when it stops, DONE with confetti.
Honest limits
Where does AI agent status stop?
What the provider, the hardware or our own roadmap does not allow yet, stated before you reserve rather than after.
Questions about AI agent status
Specific to this app. The general questions (do I need an app, does it work offline, what does a reservation cost) live on the pages that own them.
How does the display know what Claude Code is doing?
Claude Code runs hooks at set moments, and the recipe in the dashboard makes three of them POST a one-line JSON state to your app's webhook URL: working when you submit a prompt, needs_input on a notification, done when it stops. The display changes within a second of each POST.
Does the AI agent app read my code or prompts?
No. The app reads only the state word and an optional token count from the request body; the hook you install decides what to send, and the recipe sends nothing else. No outbound request is ever made to your machine.
Can it work with agents other than Claude Code?
Yes. Anything that can send an HTTP POST with a state works: other coding agents, a CI job, a long-running script. States idle, working, needs_input, done and error are understood, plus everyday aliases such as busy, waiting and finished.
What does the display show when the agent needs input?
The state changes to needs input beside the robot icon and the display plays a flash effect with INPUT, so a waiting agent is noticed from across the room. When the agent finishes, done plays confetti.
Does it count tokens?
It can. If the hook sends a token count with the state, the app keeps a daily total shown in the TOK layout, reset at midnight. The Claude Code recipe sends states only, so the count stays at 0 unless you add one.
Can Perchi react when my agent finishes?
Yes. done is a proud moment on Perchi and needs input a curious one, each with the state on the card beside the face, so Perchi tells you the agent is waiting without you watching the terminal.
More on the FAQ hub, the Perchi Matrix page and the support page.
Batch 2 - estimated 2 to 3 weeks, no payment today
Put AI agent status on your desk
Perchi Matrix is ₹4,499 incl. GST (MRP ₹4,999). No subscription needed.
No payment today. We email a checkout link when your unit is bench-tested; current estimated wait 2 to 3 weeks from the day you reserve. Cancel any time before then.