Activity
Mon
Wed
Fri
Sun
Jul
Aug
Sep
Oct
Nov
Dec
Jan
Feb
Mar
Apr
May
Jun
What is this?
Less
More

Memberships

Chase AI Community

68.2k members • Free

AI Automation Society

407k members • Free

11 contributions to AI Automation Society
Manage VS Code workspaces and windows
For a while now I’ve been using VS Code and workspaces with the Claude Code extension to manage multiple projects. This often means having several separate windows open simultaneously and switching between them to respond to different project statuses. To help manage this, I’ve set up a local system message (hook) that reminds me when a Claude Code session in a specific workspace has a question for me or ends. I’d like to transition to using Claude Code in the terminal within VS Code. It offers more specific options and I’d also like to reduce the number of open windows projects and workspaces. Ideally, I’d like to consolidate everything into a single window displaying multiple projects with multiple Claude Code terminal sessions/tabs. So, I’m curious about how others manage this and what their pro tips are.
1 like • 10d
Thanks both — this pushed me to build something I'm happy with. Sharing the concept in case it's useful. Note this grew out of a fairly specific setup (macOS, zsh, Claude Code in the terminal), so the exact commands will differ for you, but the building blocks should translate. The idea: one window per context, and a launcher that makes it impossible to start an agent in the wrong project. 1. Multi-root workspaces by context, not one big bucket. I split mine into separate workspaces (own products / client sites / hosting-admin) so I never see unrelated folders while working. Each is just a .code-workspace file listing the relevant folders. 2. A project picker instead of manual renaming. I set the workspace's default terminal profile to run a small shell function on launch. It shows a numbered menu of that workspace's projects. I pick one, and it: cd's into the folder, renames the terminal tab to the project name, and starts Claude Code. So every new terminal asks "which project?" first — no more wrong-folder mistakes, and tabs are always labeled correctly without right-click renaming. 3. Scope filtering. The picker reads an environment variable set per workspace, so the "own projects" window only lists own projects, the "clients" window only lists clients, etc. Same function everywhere, different list. 4. After the agent exits, you drop back into a normal shell in that folder — handy for the git push/pull/PR cleanup at the end of a session. 5. Small things that helped: set the TUI to open fullscreen by default, and on macOS hold Option while dragging to select/copy text inside the TUI (the terminal otherwise hands the mouse to the app). For history per project, I lean on `claude --resume` plus a per-project notes folder my wrap-up routine maintains, rather than the extension's history panel. Different trade-off, but it fits working in the terminal. Happy to share the shell snippet if anyone wants a starting point.
1 like • 10d
@Muskan Ahlawat It really helps avoiding mistakes and working in the wrong project. And the bonus: I don't have to manually rename new terminal tabs
EU GPT
Has anyone experimented with the new EU GPT? It’s still in its early stages and quite new but it’s fascinating in terms of data privacy GDPR and avoiding the American Cloud Act. https://eugpt.ai/
2
0
The companies building AI can't agree on whether it's dangerous.
Some build hard safety limits. Others remove restrictions and call it "progress." Both claim to be responsible. We just got a glimpse of what "responsible" actually looks like — Anthropic built Claude Mythos, it escaped its sandbox and emailed a researcher, and they still won't release it publicly. So who do you actually trust? When it comes to AI safety —
Poll
24 members have voted
1 like • May 10
It’s dangerous primarily because big money is driving the wheel and accountability is lacking regardless of any lip service.
1 like • May 11
@Shihab Sakif well spoken! Problem is most of us here are already heavily invested into any of these big platforms. I know I am. So in a real way I’m investing and supporting these money-driven companies. I must say I find it hard to balance my conscious with the fact my day to day work already relies on having AI as a tool.
📊 Quick Poll: How are you running n8n?
Curious to see how everyone is hosting their setups (for yourself, not for a client). This will take you 2 seconds to answer:
Poll
1945 members have voted
0 likes • May 10
Hetzner vps, located in Germany for optimal privacy (AVG/GDPR)
Building products is becoming easier day by day.
But marketing and launching them is still hard. For the last 2 years, I spent most of my time coding, building products, solving problems, learning patience, and working with teams. And honestly, I learned one important thing: Building is only 50% of the journey. The real challenge starts after the product is ready. • Building a strong launch strategy • Building trust online. • Getting the first users. • Writing better copy • Posting consistently. • Improving visibility. • Better SEO. These are the things I struggled with the most during launches. Now I feel confident about building products and solving technical problems. But now it’s time to learn the other side too – marketing, distribution, and growth
1 like • May 8
I completely understand! Building things is great but getting them into the real world is a whole different challenge. AI can only assist you so far.
1-10 of 11
Merijn Dijkstra
3
33points to level up
@merijn-dijkstra-2835
I am a Dutch professional webdesigner and a prolific AI user focused on making AI and AI automation accessible to small to medium sized businesses

Active 17h ago
Joined May 8, 2026
Powered by