Nat Friedman's Entire: a Git-native layer for AI agents
AI is writing more of our code, but the intent behind that code keeps getting lost in chats and terminals. Entire, a new platform from former GitHub CEO Nat Friedman, tackles that head-on with a Git-first approach to capturing prompts, reasoning, and tool calls next to the commits they produce.
Backed by a $60 million seed round, the goal is simple: make AI-native development traceable, reviewable, and fast - without changing how you work.
The core problem: context disappears
Traditional Git is great at tracking code diffs. It doesn't preserve why those changes happened, what an agent was asked to do, or which tools and tokens it used.
That missing context slows reviews, makes handoffs messy, and forces agents to repeat past mistakes. Entire fixes the context gap.
How Entire works: Checkpoints on Git
Entire ships an open-source CLI that extends Git with "Checkpoints." When an AI agent generates a commit, the CLI stores structured metadata linked to that commit SHA:
- Prompts and full session transcripts
- Files touched, tool calls, and token usage
- Constraints, intermediate steps, and outcomes
These Checkpoints are pushed to a dedicated, append-only branch: "entire/checkpoints/v1." Your main history stays clean, while agents and humans can query rich context without parsing noisy commit messages.
Why a separate branch instead of stuffing commit comments? Scale and clarity. Structured data belongs in a structured place, especially for multi-agent workflows where context reuse is the point.
Integrations and setup
- Works today with agents like Anthropic's Claude Code and Google's Gemini CLI.
- Planned support for others, including OpenAI Codex and the GitHub Copilot CLI.
- Install via a simple curl script, enable in any repo with "entire enable," and it hooks into pushes to sync sessions automatically.
The Entire platform: search, status, and sessions without context switching
Beyond the CLI, the service at entire.io gives you a terminal-friendly way to search commit history, check agent status, and drill into session details. You stay in your normal flow while gaining AI-specific visibility.
- Traceability: See exactly what an agent did and why.
- Faster reviews: Review intent and constraints alongside diffs.
- Better handoffs: Pick up where an agent left off without replaying chats.
- Efficiency gains: Reuse context to cut repeat errors and token spend.
- Multi-session support: Keep long-running, multi-agent work coherent.
It's open-source under MIT and free to use. Community feedback flows through GitHub Discussions and Discord for quick iteration.
Funding and industry signal
Entire raised $60 million led by Felicis Ventures with participation from Madrona, Microsoft's M12, Basis Set, 20VC, Cherry Ventures, Picus Capital, Global Founders Capital, and notable angels including Gergely Orosz, Theo Browne, Jerry Yang, Olivier Pomel, and Garry Tan.
The bet: as agents become primary code producers, traditional issue and PR workflows need durable context and agent interoperability baked in. Entire supplies that missing layer.
Why this matters for engineering teams
- Preserve human intent: Prompts, constraints, and reasoning live next to the code, not lost in chat scrollback.
- Let agents learn: Agents can reference project history to avoid rework and reduce model calls.
- Lower risk: Auditable histories help with compliance, incident analysis, and RCA.
Rollout playbook (practical and fast)
- Start with one repo and a small squad using agents for real tasks.
- Enable Checkpoints ("entire enable") and require them for any agent-generated commit.
- Update your PR template: reviewers scan the linked Checkpoint before the diff.
- Standardize agent names, branch prefixes, and tagging so searches stay clean.
- Measure impact: PR lead time, rework rate, defect leakage, and model spend per merged PR.
Get started
Install the CLI from the GitHub repo: github.com/entireio/cli. Explore the platform at entire.io and wire it into your next sprint.
If you're building out your team's AI development skills, this resource can help: AI Learning Path for Software Developers.
Your membership also unlocks:
