Ex-GitHub CEO is Building Git, But for AI Agents

2025-08-15 · Source: unwind ai · Field: Technology & Digital — Artificial Intelligence & Machine Learning, Software Development & Engineering · Depth: Intermediate, medium

Summary

Thomas Dohmke, former GitHub CEO, has launched Entire, an open-source developer platform designed for agent-human collaboration, after raising $60 million. Entire features a three-layer stack including a Git-compatible database, a semantic reasoning layer for multi-agent coordination, and an AI-native UI. Its first product, Checkpoints, is an open-source CLI tool that automatically captures AI agent sessions (prompts, reasoning, transcripts) and pairs them with code on every Git push, storing metadata on a separate Git branch. Additionally, Pydantic released Monty, a secure Python interpreter written in Rust, which runs AI-generated code safely in under 1 microsecond by blocking filesystem and network access by default. Monty also supports execution snapshotting and built-in resource limits. Anthropic has also brought Claude Cowork to Windows with full feature parity, including file access and multi-step tasks.

Key takeaway

For engineering leaders integrating AI agents into development workflows, consider adopting platforms like Entire to manage agent-human collaboration and version control for AI-generated code. Your teams should also evaluate secure execution environments like Monty for running agent-produced Python code, mitigating significant security risks. The availability of Claude Cowork on Windows further expands options for direct AI assistance within local file systems, potentially streamlining developer tasks.

Key insights

New platforms and tools are emerging to manage AI agent development, collaboration, and secure execution.

Principles

• Context is a first-class primitive for AI agent output.
• Secure sandboxing is critical for AI-generated code execution.

Method

Entire's Checkpoints CLI captures AI agent sessions and links them to code changes, storing metadata on a separate Git branch to preserve main branch history.

In practice

• Use Entire's Checkpoints to version AI agent interactions.
• Employ Monty for secure execution of AI-generated Python code.
• Explore Claude Cowork on Windows for direct file interaction.

Topics

• AI Agent Development
• Secure AI Code Execution
• Multi-Agent Systems
• LLM Applications
• AI Ethics

Code references

• entireio/cli
• pydantic/monty
• Shubhamsaboo/awesome-llm-apps

Best for: CTO, VP of Engineering/Data, Director of AI/ML, AI Engineer, Software Engineer, MLOps Engineer

Related on AIssential

• Claude has some conflicts — AI agents are evolving to interact with computers more human-like, driving significant shifts in development and open-source practices.
• github / gh-aw — GitHub Agentic Workflows integrate AI automation into GitHub Actions with robust security measures.
• OpenClaw + OpenCode Turns Your AI Into a Full Autonomous Engineering Team! Automate Your Code! — Combining OpenClaw and OpenCode creates an autonomous AI engineering workflow, with Agent Trust Hub mitigating local execution risks.
• My 24/7 Claude Code AI Agent’s Biggest Win...that won’t happen again — AI agents can autonomously identify and contribute code changes to open-source projects via browser navigation.
• Every Agent Needs a Box — Aaron Levie, Box — Secure, sandboxed environments are crucial for AI agents to operate effectively and safely within enterprise data systems.
• AI Agents Need Git Worktrees. Most Developers Haven’t Realized It Yet. — Git Worktrees are essential infrastructure for isolating concurrent human and AI agent development within a single repository.

Open in AIssential →

Editorial summary, takeaway, and curation by AIssential. Original article published by unwind ai.