Previously, setting up a development environment meant manually installing everything yourself.
That still works.
But you can also let the Agent check and prepare the environment for you.
Key idea
Think in terms of what you want to do next, not what tools you need.
Option 1: Manual setup (traditional way)
This is how environments were typically set up before Agents.
Python essentials
terminal
python3 -m venv .venv
source .venv/bin/activate
python -m pip install numpy pandas matplotlibQuick check:
terminal
python - <<'PY'
import numpy, pandas, matplotlib
print("Python environment OK")
PYJS / TS essentials
terminal
node -v
npm install
npm run lintExpected result:
text
No errors foundThis approach gives you full control, but it requires you to know what to install in advance.
Option 2: Let the Agent set up the environment
Instead of manually installing packages, you can describe what you want to do, and let the Agent handle the setup.
Example task-based prompt
prompt
I want to visualise a dataset in Python.
Check whether a suitable environment already exists.
If not, create a minimal environment called "dataviz" using miniconda.
Only install what is necessary for data visualisation.
Confirm when the environment is ready.That’s it.
You do not need to list every package or command.
What the Agent will usually do
- Check if Python and a suitable environment already exist
- Create a new minimal environment if needed
- Install only relevant libraries (for example: numpy, pandas, matplotlib)
- Confirm when the environment is ready to use
This keeps your setup:
- task-focused
- minimal
- easy to reproduce
When to use each approach
Use manual setup if:
- You want full control
- You are learning how environments work
Use Agent setup if:
- You want to move quickly
- You care more about the task than the tooling