AgentGit
Einsia AI · Coding · Productivity
AgentGit is an open-source AI agent session manager that saves your sessions as Git repositories, so the full process behind a task survives after the chat window closes. It stores the instructions, reasoning, tool calls, and decisions from agents like Claude Code, Codex, and OpenCode in a version-controlled session. Then it lets you share a read-only link, or hand the whole thing to another agent that continues from where you stopped. Ever lost hours of work because a teammate switched tools or finished for the day? AgentGit is built for exactly that moment.
About AgentGit
What Is AgentGit
AgentGit treats an agent session the way Git treats code. Every turn, prompt, tool call, and result gets committed to a repository instead of vanishing when the conversation ends. That means the work behind a task is portable. You can open it on another device, invite a teammate to approve steps, or hand it to a different agent runtime and keep going.
The problem it solves is context loss. Files hold the result. Sessions hold the process. Without the process, a new agent or teammate has to rebuild everything from scratch. AgentGit works as AI agent session version control for exactly that gap, making the session itself the shareable, versioned artifact rather than leaving the reasoning trapped inside one account on one machine.
The main limitation is that it isn't a background service. No daemon watches your disk. Sessions only enter version control when you import or push them through the hooks you install. Shared links are read-only too, and anything sensitive in a transcript is your job to strip before publishing.
Getting Started
- Run the installer with
npx -y create-agit, which puts theagitcommand in~/.local/binand wires up the setup hooks. - Sign in with
agit login --json, open the returned URL, approve access in the browser, then complete the command it gives you. - Send the setup prompt to the agent you're running (Claude Code, Codex, OpenCode, OpenClaw, Hermes, or WorkBuddy) so it reads the right skill file.
- Import or push the conversation you want to keep, which commits it to a session repository.
- Share the read-only web link, or invite someone to join the workspace and continue the work.
Product Information
A quick look at AgentGit's pricing, supported platforms, and performance.
Best for
The users, tasks, and scenarios where this tool fits best.
Users
- Developers handing off a coding session
- Small teams sharing agent work
- Anyone learning from other people's agent runs
Tasks
- Session handoff between runtimes
- Remote control of a running agent
- Archiving a finished project
Scenarios
- A teammate goes offline mid-task and you need to keep the work moving from the same context.
- You switch from your desk to your laptop and want the same agent session available on both.
- You want to show a colleague why a decision was made, not just the final output.
Key features
Sessions as Git repositories
Instead of a flat chat log, AgentGit commits each session to a repository, so prompts, reasoning, and tool calls sit under version control. You can fork a branch from an older turn and continue there when you want to take the work in a new direction. Published history is append-only. Nothing gets quietly rewritten. That keeps the record of what actually happened intact.
Cross-runtime handoff
AgentGit moves context between agents rather than locking it to one tool. Claude Code, Codex, OpenCode, OpenClaw, Hermes, and WorkBuddy all support native resume, so a session started in one can be picked up in another. It's honest about the limits. Cross-runtime conversion can lose detail, and if you hand a session to a very different runtime, agit states the losses before converting so you know what the next agent may be missing.
Remote control and collaboration
You can open the workspace from your phone or a second computer to watch an agent running on your main machine. A shareable invite link lets others join, send prompts, follow progress, and handle approvals. This is where AgentGit earns its place as an AI agent collaboration tool, turning one person's agent into a small team's shared workspace that nobody has to babysit from a single desk.
Private by default, shareable on demand
New sessions are saved privately, and you choose when to publish. A read-only link lets anyone with it read the full transcript without an account. Write access still requires authentication and a grant. Good design. That split keeps casual sharing safe without opening the door to edits.
Session hub for public runs
The public hub at /agents shows how other people's projects took shape, including instructions, attempts, tool calls, and results. It's a practical way to study prompting patterns, and it's often the fastest route to seeing how a genuinely tricky problem got solved in practice rather than in theory.
Local setup with hooks
Installing AgentGit runs agit setup --yes. That installs the hooks which save agent sessions by settling turns at the end of each user turn. There's no background daemon. Nothing runs between imports and pushes.
Pros and cons
Pros
- Open source, so you can inspect and self-host the hub instead of trusting a closed service.
- Native resume across six popular agent runtimes cuts the cost of switching tools mid-project.
- Read-only share links work without an account, which makes handing off work to a colleague frictionless.
- Append-only history means the record of a session can't be quietly rewritten.
- Remote control from another device solves the "my agent is stuck on my work laptop" problem.
Cons
- No daemon means sessions aren't tracked until you import or push them, so unsaved work can still be lost.
- Cross-runtime handoff can be lossy, and you may need to re-check the context after conversion.
- Shared transcripts can leak secrets since prompts and tool output are included, so you have to review before publishing.
- Cursor sessions import as read-only only, which limits full two-way use in that runtime.
Frequently asked questions
It saves your AI agent sessions as Git repositories, then lets you share, hand off, and resume them. The goal is keeping the process behind a task, not just the final files.
Related content
Explore related tools, skills, and articles for AgentGit.
AgentGit Alternatives
Forefront
Forefront · CodingForefront is a web platform for building with open-source AI. It lets you fine-tune leading open-source language models on your own data, evaluate how they perform, and run them through an API or export them to host yourself. Developers who want the convenience of a closed-source platform but insist on owning their models and data are the target audience here.
Startkit
StartKit.AI · CodingStartkit is a boilerplate for building AI SaaS and AI wrapper products. Think of it as an AI startup boilerplate with the boring parts already wired up: authentication, Stripe and Lemon Squeezy payments, usage limits, transactional email, and an AI API starter that talks to OpenAI, Anthropic, Groq, or Llama. You clone the repo, set your price, and start on the part of your product that people actually pay for. It's Next.js under React and Tailwind, so most of the boilerplate code already feels familiar.
Testim
Tricentis · CodingTestim is an AI-powered test automation platform for building and running end-to-end tests across web, mobile, and Salesforce applications. It leans on machine learning to keep tests stable when an interface changes, so teams spend less time fixing broken selectors. Not bad for an automated testing tool you can start using today. You create tests by recording actions in a browser, then optionally add JavaScript when you need more control. It's a solid pick for busy QA teams.