@Kevin-0, it should work now, but only if you have a paid ChatGPT "plus" subscription or greater.

To add the TRIGGERcmd MCP server (https://www.triggercmd.com/mcp) to your ChatGPT account:

Enable development mode here: https://chatgpt.com/#settings/Connectors/Advanced
8704bf08-29d3-4e8b-9382-599dd8e2f552-image.png

Go here: https://chatgpt.com/#settings/Connectors
237586cd-f0c4-4b06-9744-6cfeedef3867-image.png

Click "Create app" and set it up like this:
2a885474-3eb8-4734-80f7-9c37750770e3-image.png

I paid for the ChatGPT plus upgrade and tested it:

121d4d94-e7af-4eda-966a-7d8e1af29114-image.png

EDIT: I submitted the MCP server to OpenAI to be included in their app store. That should solve the issue I've seen where ChatGPT runs the command twice. If they reject the app I'll do something to debounce it.

Thanks @kellanist. I'll take a look.

I hoped Microsoft would whitelist apps signed by my code signing certificate by now. Maybe I'm missing something.

If it's new, I think it must be because I recently renewed my certificate.

EDIT: I submitted the agent to Microsoft just now, so hopefully they'll whitelist it soon. Otherwise I just have to wait for its reputation to build as users install it.

0cde63ea-3d0e-4b4a-b6e0-13c398ae26ef-image.png

This is what I saw when I submitted the agent installer:

"The SmartScreen warning that you reported indicates that the application and/or certificate used to sign it do not yet have a reputation established in our system. Applications without a known reputation will show warnings when downloaded or installed. This does not prevent users from still clicking through to download or run the application. To do so, they can select "More info" -> "Run anyway" option within SmartScreen prompt window.

Reputation is established by how your download is used by Windows, Edge, Internet Explorer users and the SmartScreen® Service intelligence algorithms. Downloads are assigned a reputation rating based on many criteria, such as download traffic, download history, past anti-virus results and URL reputation.  This reputation may be based on the downloaded program or can also be assigned to the publisher, based on digital certificate information.

Once your signing certificate has gained reputation in our system, all applications or releases signed with your certificate should have warn-free experience"

@mega_big, on a Raspberry Pi, assuming you installed the agent using these instructions, your commands are in this file:

/root/.TRIGGERcmdData/commands.json

You'll see the default commands in there.

You can run these commands to edit it:

sudo su - cd /root/.TRIGGERcmdData nano commands.json

While the agent is running, it will detect changes in that commands.json file and create/remove triggers for those commands in your TRIGGERcmd account. If you mess up the json at all (an extra comma even), the agent will restore the last backup it took. You'll see that in your debug.log.

If you don't want to deal with editing text, you could create a command on your Windows PC using the "GUI Command Editor" then copy/paste the json text of that command from the "Text Command Editor" into your commands.json file.

@JoshuaJSlone, that's a known issue with the windows timeout command. It doesn't work in contexts that don't have an interactive terminal, like when running it via TRIGGERcmd. Here's one method that does work:

start /wait timeout 3

I've also added delays with the ping command like this:

ping 127.0.0.1 -n 2