Our first ship at Entire, Checkpoints is a new primitive that automatically captures agent context as first-class, versioned data in Git. Try today in Claude and Google Gemini CLI. Coming to all agents soon.

@claudeai @GeminiApp made with @Remotion ❤️

@EntireHQ @claudeai @GeminiApp congrats on the launch! we at @WorkOS would love to help you quickly ship this to enterprise with SAML, SCIM, audit logs, EKM. and more 🫡

@EntireHQ @claudeai @GeminiApp Congrats Thomas! Can’t wait to try it later today!

@EntireHQ @claudeai @GeminiApp ❤️

@EntireHQ @claudeai @GeminiApp this is amazing! thinking of the ecosystem of products that will be built atop this. very exciting product! congrats on the launch!

@EntireHQ @claudeai @GeminiApp What problem does capturing agent context as versioned data solve that existing solutions can't. Does it improve collaboration or something else. How does this change workflows

@EntireHQ @claudeai @GeminiApp Wow version control gets extended that’s clever and soon enough we probably won’t live without it. Very cool

@EntireHQ @claudeai @GeminiApp Check MemoV

@EntireHQ @claudeai @GeminiApp Congrats on the ship!

@EntireHQ @claudeai @GeminiApp this is huge, congrats with the ship!

@EntireHQ @claudeai @GeminiApp Very cool, congrats on the launch 🚀

@EntireHQ @claudeai @GeminiApp Ha, I was wanting something like this just yesterday (git checkout but for agents) and started prompting away... then I find this the next day. This is great, it's exactly what I'm looking for. I'll be a heavy user at my day job.

@EntireHQ @claudeai @GeminiApp Keep the feedback coming. Let’s make it happen.

@EntireHQ @claudeai @GeminiApp The % of "my code" that is hand-written is declining weekly. Adapt or die. 90% of the productivity bottleneck is code review and 10% context management. While I won't be able to use the UI due to our data security policy, code review tooling is what I'll be watching most closely

@EntireHQ @claudeai @GeminiApp I'm very curious to see what your thoughts are on the future of CR. Me & team still strongly believe in line-by-line review, but this is becoming increasingly impossible with the volume of work that can be produced by Claude & agents.

@EntireHQ @claudeai @GeminiApp You will review code as long as you need to for security, quality, and ultimately trust. You will stop when those are good enough when done by an agent and everybody around you has already moved on.

@EntireHQ @claudeai @GeminiApp AI can flounder without a detailed technical spec. Without human review, it will be difficult to build enough understanding to prompt a good spec. Do you believe models improve to the point where increasingly high-level tech specs will be needed--eventually only business sense?

@EntireHQ @claudeai @GeminiApp Yes, I do. Compare Copilot in 2021 to Claude Code or Codex in early 2026, and extrapolate that into the future, and tell me why you wouldn’t.

@EntireHQ @claudeai @GeminiApp I was talking about how we needed this on standup today

@EntireHQ @claudeai @GeminiApp So exciting. One question, how do you see Entire integrating with existing GitHub workflows as AI agents become more central?

Lots of ideas here. For example, the code review agent can leverage the checkpoints for context on what the developer thought during code generation. Of, if you have a deployment gate with manual approval, the approver can get a perspective on what has changed beyond just the commit message or PR description. Intent flowing through the SDLC will be incredibly powerful.

@EntireHQ @claudeai @GeminiApp can’t wait to see this in action. Do you imagine this primarily as deep GitHub integrations, or something that stays platform-agnostic?

@EntireHQ @claudeai @GeminiApp Oh, definitely platform-agnostic.

@EntireHQ @claudeai @GeminiApp Oh good, a permanent history of all the slurs and insults I hurl at Claude

@EntireHQ @claudeai @GeminiApp So the real target audience of Entire seems to be coding/reviewing agents, not human developers. We’re entering a phase where tools are built for AI first and humans second. Interesting times to be in.

@EntireHQ @claudeai @GeminiApp @theo u probably saw this but pinging just bc

@EntireHQ @claudeai @GeminiApp need to try this

@EntireHQ @claudeai @GeminiApp Free the Codex!!

@EntireHQ @claudeai @GeminiApp We dug into the Entire CLI to see what actually shipped beyond the funding headline. Git-native agent session capture is a clean first product, but the $300M valuation is pricing in a platform that doesn't exist yet. Full diligence breakdown here

@EntireHQ @claudeai @GeminiApp I like that this is now a possibility, been waiting for something like this since Zed brought up DeltaDB

@EntireHQ @claudeai @GeminiApp This is actually clever - versioning agent state in Git gives you proper rollback and branching for agent workflows. Are you storing the full context or just key decision points? Curious how you handle merge conflicts when agents diverge from a checkpoint.

@EntireHQ @claudeai @GeminiApp Love it. Will it cover all surfaces too? Not just clis?

@EntireHQ @claudeai @GeminiApp Can this be triggered directly from a Gitlab pipeline? Would also be cool if the agents can map changes to checkpoints on a business roadmap

When the GitHub CEO is posting about multi-model AI integrations, you know the developer tooling landscape is shifting fast. The future isn't about picking one AI provider — it's about orchestrating all of them. The best developer tools will be model-agnostic. Entire seems to get this.

@EntireHQ @claudeai @GeminiApp Congrats on the launch! This is neat. For better redaction (patterns not entropy), passive capture, curstor support and encrypted share etc I think you should look at Contrail. Think you should also look at local first as an option.

@EntireHQ @claudeai @GeminiApp Enabled it in a few repos now. One thing I am missing: How do I best use what is captured? I'd imagine a workflow similar to @kieranklaassen 's compound engineering where an agent can go fetch and search the context based on a line/code block in the repo when working on it.

@EntireHQ @claudeai @GeminiApp curious how granular the checkpoints get - per task or per major decision?

@EntireHQ @claudeai @GeminiApp V stoked

@EntireHQ @claudeai @GeminiApp that's cool, glad to see new primitives like Checkpoints making agent data more accessible

@EntireHQ @claudeai @GeminiApp To be honest, Git feels increasingly legacy, and any innovation built around it may eventually need to evolve beyond the current technical foundation.

@EntireHQ @claudeai @GeminiApp Any suggestions about how to structure/a workflow for the prompts? It sounds like the prompts themselves should be carefully constructed to have any value for other people or agents - now or in the future? Btw. have hard time setting this up in Windows (only on wsl for now?).