First Useful Task

Station 9 — First win. Time: 20 min. Pass when you have copied or saved one useful output.

What this is

This is your first real win. You point Hermes at one genuine task and walk away with something you can use. The output is your proof — and the habit of verify, then save is what you are really learning.

Why it matters

Setup is not the win. A useful result is. Producing one real output today is what makes the whole session stick and gives you a reason to open Hermes again tomorrow.

What to do

Use the task you named in Station 1. Here is a full worked example you can copy.

1. Ask with a clear, bounded prompt.

I am new to Hermes. Help me complete one useful task today.
My task is: turn these messy meeting notes into a clean action plan.
Notes: [paste your rough notes]
Ask only the minimum questions needed, then produce a clear action plan
with owners and next steps.

2. Answer Hermes' questions. A good agent asks a couple of sharp questions first — who owns what, what the deadline is — then works. Answer briefly; you do not need to over-explain.

3. Read what it produced. You should get something structured, for example:

Action plan
- [ ] Draft the budget summary — you — by Thursday
- [ ] Book the venue — Aisha — this week
- [ ] Send recap email — you — today
Open questions: final headcount?

4. Verify before you trust it. Skim for anything wrong or invented. Ask Hermes to fix one thing if needed: "Split the first item into two steps and add a due date."

5. Save or copy the output. That saved result is your proof for today.

Two commands that help right after:

• /status — a quick recap of what happened in this session.
• /usage — how many tokens it used and roughly what it cost.

Not into meeting notes? Two other strong first tasks:

Draft a polite email declining a meeting and proposing two other times.
Context: [who it is, why, your availability]

Summarise this document into 5 bullets and 3 next actions.
Document: [paste text or key points]

Common mistakes

• Asking a vague question instead of naming a real, bounded task.
• Trusting the first output without skimming it for mistakes.
• Not saving the result, so there is nothing to show afterwards.
• Letting Hermes ask endless questions — tell it to keep them minimal.

Next step

Go to the Slash Command Toolkit.