# Pharos is taking you through this one.
# Free, and it runs right here. No signup.
$ learn Continue
Learn Continue in 5 minutes.
Continue is an open-source coding assistant for VS Code and JetBrains that you configure yourself: your models, your context, your rules. The lesson drills its four modes, feeding it context with @-mentions, and understanding what the config file controls.
# 5 minutes, no install
# interactive, not a video
# Lesson not loading? Open it in a new tab. It is the same free lesson either way.
Chat, Edit, Autocomplete, or Agent?
Continue is four assistants sharing one extension, and each is tuned for a different size of task. Reaching for the wrong one is the main way it disappoints.
Ask with context
A conversation that can see your code when you point it there: @-mention files, docs, or the whole codebase. For understanding and planning, before anything gets written.
Change the selection
Highlight code, describe the change, review the diff. Scoped to exactly what you selected, which makes it the safest of the four modes and the one to reach for first.
Hand over a task
Agent mode works across files and can run tools to complete a goal you describe. Save it for changes too broad for a selection, and keep the goal bounded.
What you will do in the lesson
You answer and try things. Nobody demonstrates at you for five minutes.
- Match the mode to the task. You'll sort a stack of everyday coding jobs into Chat, Edit, Autocomplete, and Agent, and see where the boundaries actually sit.
- Point instead of paste. You'll practice @-mentioning the right files and docs in Chat, and compare answers grounded in your code against answers from a model left guessing.
- Run an Edit and read the diff. You'll transform a selection from a plain-English instruction, then review the proposed change line by line before accepting it.
- Read a config with fresh eyes. You'll walk through a Continue configuration and identify which model serves autocomplete, which serves chat, and why teams split them.
- Decide when Agent earns its risk. You'll take one multi-file task, brief the agent tightly, and note where you'd want to check its work.
Before you start
Why configure my own assistant instead of using a preassembled one?
Control and fit. You pick which models power each mode, which can mean a fast local model for autocomplete and a stronger one for chat, and you can point context at your team's own docs and code. Preassembled tools choose all of that for you and keep the choices hidden.
Do I need VS Code open to take this lesson?
No. The lesson is interactive and runs in this page, with nothing to install and no account. Installing the extension comes later, if you decide it's worth it; Continue itself is free and open source.
Can Continue run fully local, without sending code anywhere?
Yes, that's one of its main draws. Point it at models running on your own machine through a local server and your code never leaves it. Plenty of people mix: local model for completions, hosted model for hard questions.
How is this different from Cline? Both are open-source extensions.
They emphasize different halves of the job. Cline is agent-first: give it a task and it plans, edits, and runs commands. Continue is assistant-first: autocomplete, chat, and scoped edits across your whole day, with an agent mode included. Heavy delegators lean Cline; people who want AI woven through ordinary editing lean Continue.
Other tools people learn next
Same format, same five minutes, still free.
Learners from these organizations are already learning on 
