Introducing CodeGrid!
The terminal workspace for AI coding agents.
Run dozens of Claude Code another agent sessions in multiple infinite canvases. Drag, resize, and zoom each pane. Status indicators show what every agent is doing, across all workspaces.
Native macOS. ~10MB. Open source.
https://t.co/iZkEUWNZFe
1/ 🧵
We just shipped something we're genuinely excited about: a private, every-model AI coding agent inside CodeGrid, powered by @AskVenice.
You can now run your own Venice Powered Terminal inside @CodeGridDev
Here's how it works 👇
GM! If there's something that you want added to the @CodeGridDev app today, let me know what it is and I'll try to complete it by end of the day! (If it makes sense, of course)
Staking $GRID is now live on @base 🟦
Think of it as a subscription you don't pay for: lock $GRID, unlock @CodeGridDev Pro inside the app. You never spend your tokens...your principal stays 100% yours.
Here's how it works 👇
Going over all of the @CodeGridDev user feedback from last week.
Everyone that sent me feedback, just know that we are actually taking it into consideration and making changes to what needs changing.
As usual, tons of product improvements will be implemented this week along with some surprises.
Over the past year, my team and I have built some products I'm genuinely proud of. Two favorites are now live and run by different teams:
→ https://t.co/efoV4ZspIJ : an AI agent that writes AEO optimized blog articles from real data and analysis
→ https://t.co/7lEOmn8lbr : an AI agent that analyzes all your Instagram videos and images, learns your page, and sends contextual replies to every comment
While building these, my devs and I kept hitting the same development workflow problems. So we started building CodeGrid on the side to solve them.
It was built with the indie developer in mind. Now we've got a dedicated team working on @CodeGridDev full-time, fixing the exact issues we ran into ourselves.
had a call with the @bankrbot team last week.
they were excited about what we are building with @CodeGridDev
I have started multiple successful companies over the past 10 years and I know legit hardworking people when I speak to them.
very impressed with all @igoryuzo@0xDeployer and @Dannyhbrown have done until now.
quick rundown of what came up 👇
It doesn't merge them into one brain.. it isolates them.
Each project = its own workspace of agents
Switching projects never kills them, they keep running in the background
Sessions persist + resume on restart (--continue), no cold starts
And within a project, the agent-bus lets agents read & message each other directly
The win isn't shared memory. It's that nothing gets lost when you switch.
High slippage on Uniswap comes down to how the pool works, not the project. Trades happen against a liquidity pool using a fixed pricing formula, so larger orders move the price more... especially when liquidity is still building. This is completely standard for newer tokens and improves as more liquidity gets added.
To reduce it:
Trade in smaller amounts
Use a limit order so you set your max price instead of accepting whatever the market gives you
That's it ...the impact you saw is a function of order size vs. pool depth, nothing more.
MAJOR!
GRID now has a live, public treasury...every WETH and $GRID fee claimed from the Doppler pool is visible on-chain, with a named wallet, a written policy, and a terminal UI you can actually read.
Learned a lot about @bankrbot in the past 48 hours! what an amazing platform @igoryuzo@0xDeployer
https://t.co/hW0Mx9FLLm
Just opened a PR to add @CodeGridDev as an OpenClaw / @bankrbot
Skill: https://t.co/nDoj0IQyB0
Once merged, any agent can install the codegrid skill and instantly know how to drive a multi-agent canvas, spawn panes, message sibling agents, coordinate work across Claude, Codex, Gemini, Cursor, Grok, and shells.
@0xDeployer@igoryuzo 🙏
For anyone that was curious and wanted a quick explainer video on what @CodeGridDev is and how it works...here you go
I'm also answering all questions. Anything you have, please comment below!
Well, I think Claude Code's app personally has way too many things that are not development related. Plus, you can only use Claude and no other agents. I typically pair Claude and Codex and delegate different tasks depending on what I'm doing.
Code quality only becomes an issue if you don't guide the agents properly and give them non-conflicting tasks.
However, if you use the Agent Bridge MCP ... You can have Claude or Codex delegate tasks to other agents and they will all kind of work together.
Open to hearing your thoughts after trying the product.