Ask for it
Say what the control should be and what it should do.Give me a button that runs the report now.
Add a box where I paste a link, then summarize whatever is at that link.
Let me upload the CSV and clean it up.The agent builds the control into your Triggers tab. That page is a real app, so a manual trigger can be a button, a text box, a file upload, a dropdown, or a short form. Ask for whatever fits the job.
Ask for the inputs you need
The difference between a useful control and a frustrating one is usually the inputs. Think about what changes from run to run and ask for a field for each one.Add a date range picker so I can rerun last month instead of just today.
Let me pick which customer before it runs.You can also ask for output. A run that ends with nothing on screen leaves you wondering whether it worked.
Show me what it found when it finishes, and tell me if it fails.
Use it to test your other triggers
Every trigger on an automation runs the same code. So a manual button is how you check a scheduled task without waiting until tomorrow, and how you check a webhook without asking another service to call you. Ask for one early. Keep it after everything works.Add a run-now button so I can test this before the schedule kicks in.
Worth knowing
- Manual runs happen right away. Your workspace is already awake when you are looking at it, so you skip the wake delay the other triggers have.
- Watch it happen. Since you are sitting there, this is the best moment to catch a wrong assumption. Read the output on the first few runs.
- You can stack triggers. A manual button does not replace your schedule or your webhook. Keep all three on the same automation.
Change it later
Move the button to the top of the page.
Add a confirmation step before it sends anything.
The dropdown should list only active projects.