/
Menu

Manage connected agents

See your connected agents and when they were last used, mark a key an agent has seen, and disconnect an agent without a passkey check.

The Agents page lists every connected agent. This page helps you review them, mark keys an agent has seen, and disconnect one.

#See your agents

Open Agents. Under Connected agents each agent shows:

  • Its name, and a badge: "Signed in via" the app's name with "unverified app" for an agent that signed in through the browser, or Manual token.
  • Connected with a date, and "last used" with a date, or "not used yet".
  • The projects it may ask about, as badges.

The name of a signed-in app is the app's own claim. Check the agent name you chose when you connected it.

Everything an agent does is recorded: when it connects or disconnects, each request, each approval and denial, and each playbook step it reports. You can see it in Settings, under Activity. If an agent's sign-in is reused or shows up in a different app, StackHangar disconnects that agent and logs why.

#Disconnect an agent

Disconnecting needs a signed-in session but no passkey check, so it is fast in an emergency.

  1. On Agents, choose Disconnect next to the agent.
  2. Choose Disconnect now, or Cancel.

The agent loses access at once, and any request or proposal it made that you haven't answered is cancelled. Nothing it was already given changes: references stay valid, and a value you showed or pasted somewhere is out of StackHangar's hands.

To connect an agent again, see Connect an agent.

#Keys an agent has seen

If an agent created a key itself, for example by running a provider's command-line tool, or read it from a file, it saw the value. Treat that key as exposed.

StackHangar can't tell what an agent saw, so you mark it:

  1. Open the item and edit it. The box appears on API key and database items.
  2. Tick An AI agent created or saw this value.
  3. Under Which agent?, pick one of your connected agents or type a name. If you leave it blank, StackHangar uses "An AI agent". When you have exactly one agent you named yourself, it is filled in for you.
  4. Save.

You can also tick An AI agent wrote this file or could read it once when you paste a .env file on the import page.

The item then shows "Seen by an AI agent" and is listed to rotate first. The mark is encrypted with the item. Rotating the key and saving a new value clears the mark. If the agent can read where the new value goes, such as a .env or a deploy target, mark it again.

Keys you add through an approval link are not in this group. You type the value into StackHangar, and the agent never sees it.

See Respond to a leaked key.