Why teams replace assembled cloud services with KERNSTACK Hosted Git
Teams otherwise keep code in a general git host and wire deploy keys, mirrors, and branch conventions by hand — then discover their agents cannot participate cleanly. On KERNSTACK the repository is part of the project: clone, branch, and deploy speak the same names.
Working with the repository
kernstack git clone brings down a console project with auth and large-object filtering already configured, pushing an initial commit with a root stack.json if the remote is empty. kernstack git branch lists remote branches including agent/
Why git is a platform feature here
Because agents are first-class authors. Agent demos live on agent branches that map to agent environments; hosted git is the medium that lets a person pull the agent’s exact state, review it with ordinary tools, and promote it to a feature branch. Deploys use their own bundles, so repository compaction in the console never disturbs what is already running.
- The console’s Kernstack Git panel shows storage, branches, tags, and last compaction
- A project created in the console gets its initial stack.json commit on first clone-push
- Keeping agent work is one offered step: kernstack use proposes feature/
when you leave an agent branch
SDKs & Language Integration
Every KERNSTACK environment vendors typed SDK packages into .kernstack/ so both engineers and coding agents in Agent Sandboxes can call Hosted Git without external credentials:
- Go SDK:
kernstack CLI and console (no SDK client) - Node.js SDK:
kernstack-sdk - Python SDK:
kernstack_sdk - PHP SDK:
kernstack/sdk (Composer)
Frequently asked questions
No — a local repository can plan and deploy. Hosted git earns its place through the agent workflow: agent branches, demo environments, and clean checkout of agent work are built on it, and @agent code mode requires it.
The repository is a git remote; mirroring arrangements are ordinary git. What matters is that agent branches and deploy targets speak hosted-git names, so most teams keep the canonical project repository here.
Yes. Clones configure large-object handling automatically, and asset-heavy projects store their bulk in buckets as usual — git carries code and declarations, buckets carry the bytes.
Related capabilities, architectures, and comparisons
Ship your stack on the Software Lifecycle Cloud.
Closed preview on dedicated European infrastructure. Bring a repository or start from a blueprint.