Unpopular opinion: VS Code is holding back AI coding. I’ve been spinning up multiple Cline instances across tmux panes and terminal tabs. Each agent has its own isolated state and can run a different task, branch, or idea in parallel. They just keep going while I focus on something... else, no context collisions, no babysitting a single chat thread. You can pipe input, chain output, or run them headless in CI/CD. It fits into the workflow I already use in the terminal. Getting started literally took me 30 seconds: npm install -g cline AI is finally just another process in my workflow, instead of a panel or plugin.show more

is so much better for me

This is actually a solid take. When AI lives inside an editor panel, it still feels like a feature. When it runs in the terminal like any other process, it becomes part of the workflow.

I'm very impressed with their terminal UI. As a developer I love working with the terminal. Each agent having a separate state is an added bonus to speed!

Totally. The UI is nice, but the isolated state is the real win. It maps cleanly to how we already think about branches/tasks. One idea, one agent. No weird cross-talk. That’s what makes parallel work actually viable.

This is very impressive!

Indeed! thanks

Interesting! I am also looking for an alternative for VSCode since they are clearly losing the AI race with their shitty copilot XP. I will try Cline out 👀

@SquidCorp_ink To be fair, VS Code is great at what it was designed for. I just don’t think AI fits neatly into the "sidebar assistant" model. Running agents in parallel, headless, or inside CI feels way more natural once you try it.

Indeed, it is good at what it is designed for. The sidebar looks like a hacky design fix just to say - Hey, we also have AI here. But the UX is bad. It's slow. Try to interrupt if it takes too much time... 🚫

🙌

🫡

not so sure, many people still prefer vscode extensions than cli , at @A2AxDev we support both for managing and purchasing skills, we're seeing traction in the extensions mroe than expected

@A2AxDev Totally fair. I don’t think CLI replaces extensions for everyone. A lot of devs live in VS Code all day and want AI right there. My point is more that the process model unlocks workflows extensions can’t, especially around parallelism and CI. Different tools for different habits.

@A2AxDev i think we still need to see the final boss ui that will drive ai forward, probably something we can't imagine as we're still just adapting our ape tools

Treating AI like npm, git, or docker instead of a sidebar is the right direction. If it can be piped, scripted, and killed like anything else, it becomes part of the system. Great work @cline

@cline composability is the key

Impressive my friend

vscode can spin off multiple sessions right from within the editor? It can spin them off to a background copilot-cli process. It can spin them off to the cloud if that's what you want and your code is on github. It has subagents, skill support, UI components in chat...

That isolation for parallel tasks sounds key for agent autonomy.

i can do the same from the terminal in VS code... as a bonus i can see all my active sessions, review diffs in real-time, work on documents while the agents are working, and have acces simultaneously to claude, gemini, codex (including all the model versions)

VS Code devs are intentionally delaying ACP implementation. Technically, you no longer need any code editor to vibe code.

I wish it was open source but there is GitHub Copilot CLI which you can run multiple instances of.

Once AI runs like background processes instead of chat windows, it actually feels native to the workflow.

I’m using Monaco editor with a web front end

Agree on terminal-first. But the real unlock isn't parallelism - it's composability. Hooks that intercept tool calls before execution, rules that load on-demand, git worktrees per agent branch. When the agent is just another process, your entire existing toolchain works with it naturally - pipes, scripts, CI, everything. Extensions can't compose like that.

This is such a nice idea! Thanks for introducing me to Cline

While i use opencode, and pi in scripts , i still prefer to have vscode open all the time. Usually few of them. you can run multiple parallel sessions in one, although the session specific branch would be nice. With subagents it’s easy to get to a few hour sessions. Works fine.

This is the edge most devs miss. Real productivity isn’t happening in bloated IDEs.

the parallel tmux setup is honestly where it clicked for me too. having 3 agents working on different parts of the same project at once feels wild compared to waiting on one chat thread in vscode

Vscode is holding back coding since way before LLM tbh

What’s the advantage of cline vs just running claude code or codex?

Another Impressive CLI! Will give it a try

For me it's opencode

💯 Completely agree! Tmux and an CLI agent means more interfacing through pure text and definitely beats an IDE

Parallel task execution in Cline is really a big time saver

running 8 parallel tmux tracks is the unlock. each agent gets its own worktree, no file conflicts, no context bleed. woke up to 29 builds, 238 tasks done, $0.42 spent. the bottleneck was never compute.

This was missing: open-source, model-agnostic, editor-agnostic. 👌

Nice setup. Isolation + headless runs feel much closer to how dev tooling naturally scales.

CLI agents > sidebar ghosts

And you end up with an absolute clusterfuck of a codebase. Stop promoting retarded ideas

I don’t see the hype. @opencode can do this and even “opencode web” gives this to you without the need to open up multiple terminal tabs. Even Kimi code has a web GUI with isolation

This looks nice