@itsclelia: I have one big problem with agentic engineering: I want agents to operate autonomously, but I also want granular, rever…
Summary
I have one big problem with agentic engineering: I want agents to operate autonomously, but I also want granular, reversible control over every change they make. I could solve this by committing every intermediate step to Git, but that would completely pollute my repo history. So I built 𝗮𝗴𝗴𝗶𝘁: a Git-like CLI for local and remote (S3-backed) agent artifact storage, written in Rust . With aggit, my agents can stash intermediate work, create branches safely, restore previous states, and back
View Cached Full Text
Cached at: 05/11/26, 10:38 AM
I have one big problem with agentic engineering: I want agents to operate autonomously, but I also want granular, reversible control over every change they make. I could solve this by committing every intermediate step to Git, but that would completely pollute my repo history. So I built 𝗮𝗴𝗴𝗶𝘁: a Git-like CLI for local and remote (S3-backed) agent artifact storage, written in Rust . With aggit, my agents can stash intermediate work, create branches safely, restore previous states, and back up artifacts to my @RustFSofficial server with multi-instance replication support, all without touching the main Git repository. Once the results are ready, I can cleanly push the final changes to GitHub. Learn more → http://aggit.dev Install as an agent skill → 𝘯𝘱𝘹 𝘴𝘬𝘪𝘭𝘭𝘴 𝘢𝘥𝘥 𝘈𝘴𝘵𝘳𝘢𝘉𝘦𝘳𝘵/𝘢𝘨𝘨𝘪𝘵 Leave a → http://github.com/AstraBert/aggit
aggit | S3-backed, git-versioned object storage
Source: https://aggit.dev/ Rust CLI + S3
S3-backed, git-versioned object storage for agents. Track local files as compressed objects, branch safely, and sync reachable history to S3-compatible remotes.
$ aggit init .
$ aggit author -n "Jane Doe" -e "[email protected]"
$ aggit add notes.txt
$ aggit commit -m "initial commit"
$ aggit origin create prod -e https://s3.example.com -s *** -k *** -r us-east-1
$ aggit push prod
Core Capabilities
Git-like Local Engine
`blob`, `tree`, and `commit` objects are stored under `.aggit/objects` with branch-aware refs and index.
Branching + Recovery
Switch branches with clean-state checks, restore working trees from committed snapshots, and inspect history objects.
S3 Sync
Push only reachable objects beyond remote head; clone remotes back into `.aggit` and materialize files locally.
How It Works
`aggit` keeps a minimal Git-like store inside `.aggit/`: file contents are hashed into compressed objects, index entries are tracked per branch, and commits link trees plus parent pointers. On push, it computes objects reachable from local head and uploads only what the remote bucket is missing, alongside branch head and index. Clone performs the inverse: downloads `.aggit` data from S3 and restores the working tree from the selected branch commit.
Similar Articles
Grit: Rewriting Git in Rust with Agents
The article details Grit, a new Rust reimplementation of Git that passes over 99% of the Git test suite, created using AI agents. It aims to provide a library-based, memory-safe alternative to the original Git.
@yoheinakajima: my current @activegraphai inspired approach to a modular repo-centric agent operating system
Yohei Nakajima outlines a repo-centric modular architecture for agentic work, where the durable unit is a governed project repository rather than the model or conversation, with agents acting as replaceable execution components over persistent state.
We are treating AI like a magic trick instead of software, and it’s making agents unmaintainable.
The article argues that current AI agent frameworks treat agents as black boxes, making them unmaintainable, and proposes a Git-native architecture (Lyzr GitAgent, OpenGAP) where agent logic is version-controlled as flat files with pull requests for rollback and auditability.
@akshay_pachaar: Stanford researchers did it again. They just built the agent-native version of Git. When an agent works on a longer tas…
Stanford researchers released Shepherd, a runtime layer that acts like Git for AI agent runs, recording typed events and enabling copy-on-write forks so agents can revert to previous states and reuse KV cache. Early tests show improved pass rates for multi-agent cooperative tasks.
I built agentwerk, a tiny Rust crate for scaling agent collaboration focusing on getting work done
The author introduces agentwerk, a lightweight Rust crate designed for scaling agent collaboration through a ticket system, aiming to provide a simpler alternative to bloated frameworks like LangChain or AutoGen for human-in-the-loop-free pipelines.