AI Decision System
Replayable, offline-verifiable records for AI-assisted go/no-go decisions: experiment promotions, typed model answers and human-approved actions.
The company's own cloud on one machine, installed and operated by an agent.
Developer toolsJavaScript · ShellGitHub @EstocaOnlineOwner unconfirmed
Have you tried it? One tap is enough. No sign-in needed.
No updates detected yet. GitHub releases will appear here.
Owners of public repositories under a personal GitHub account can sign in to verify ownership and manage the introduction themselves. No administrator approval is needed.
Sign in, generate a verification file, and commit it to this repository’s default branch within 30 minutes. Files on other branches or forks do not count. Then check the file below to manage the project without administrator approval. Keep the file to retain your access.
If needed, you will go to sign-in. After signing in, return here and press your chosen verification button again.
Run /nomorevibe verify in your project folder to verify ownership and manage the introduction yourself.
Verify in one minuteAsk about other verification options
The company's cloud on one machine: SSO, git with push-to-deploy, shared Postgres/Redis/S3, backup and an optional operator agent. An agent installs it; an agent puts apps on the air.
Taken from the repository README
Loading the full README.
Read the full README ↗The company's own cloud on one machine: an agent installs it, and an agent puts apps on the air in it.
One box runs what a small company needs around its internal apps: single sign-on (Authentik), Git with push-to-deploy (Gitea and Actions), shared Postgres and Redis, a portal, an MCP gateway for AI clients and a nightly backup, all behind one sign-in and under one contract an agent can read. The box can be a Mac with Docker Desktop, a VM on a laptop or a server in the office. Open source, MIT.
Documentation: https://inhousecloud.estoca.com.br/ (built from docs/).
Sibling project: InHouseOS is the people-and-work half (a personal LLM agent per person, a second brain, a shared tracker, a company compass). It runs on top of this cloud as ordinary apps and is adopted separately.
On a Mac with Docker Desktop, directly (15 to 20 GiB free; no sudo):
git clone https://github.com/Estoca/inhousecloud.git ~/src/inhousecloud && cd ~/src/inhousecloud
./install.sh --non-interactive --config bootstrap/config.example.yaml --install-dir ~/inhousecloud
Or in a throwaway VM, on a Mac with Homebrew and no Docker (or skip the VM on any Ubuntu box with Docker):
brew install lima
limactl start --tty=false --name ihc --cpus 2 --memory 4 --disk 30 template:docker
limactl shell ihc
# inside the VM:
cd ~ && git clone https://github.com/Estoca/inhousecloud.git && cd inhousecloud
sudo install -d -o $(id -un) -g $(id -gn) /opt/inhousecloud
./install.sh --non-interactive --config bootstrap/config.example.yaml
About a minute for the VM and five for the install. The install ends with seven day-zero steps, with the literal commands: open http://portal.cloud.localhost and sign in, turn on MFA, copy the backup key off the machine, put a first app on the air with scripts/ihc-app.sh new <name> --owner <email> --push (about 30 seconds later behind the same sign-in), and stop or remove a test install (docs/cleanup.md). Both paths, with the times measured: docs/try-it.md.
Or ask a coding agent. Below, Claude Code gets one sentence in Portuguese ("create an internal expense log app on our company cloud and put it on the air") and ssh to a box it has never seen; recordings and how they were made are in docs/demo/.
Each of these ran on a clean machine and is written down with the commands, the times and what broke, in docs/first-run.md:
new to the air by a push to main (app loop);private ingress over Tailscale with the installation's own CA (private ingress run);ihc doctor against failures caused on purpose (doctor run), and the operator agent in a conversation (operator run).The core (proxy, Authentik, Gitea with its runners, shared Postgres and Redis, the portal, the MCP gateway, the backup) is always on. Each profile adds:
| Profile | Adds |
|---|---|
storage | Garage, S3-compatible object storage: a bucket and key per app |
ops | Portainer, Glances, a web terminal and a file browser, for admins only behind the SSO |
agent | an operator agent that answers "why is it down" in Slack or Telegram, read only by default, inside a written contract |
Start at AGENTS.md: it routes an install, a new app or "why is it down" to the contract and the playbook (docs/llm-bootstrap.md, docs/agent-setup-contract.md, docs/operator-contract.md). Off the box, an agent reads the cloud's health through the portal API, and a person's AI client uses it through the MCP gateway.
Usable on one box, with local ingress (on the machine itself) or private ingress (Tailscale, tested end to end). The portal has people and groups and a read-only status API for agents; the MCP gateway serves read-only tools, and every new app is born with one. Next: write tools and an audit log in the gateway, restoring one app on its own, and the off-site backup copy. public ingress, for a cloud on the open internet, comes after that: the installer has the modes, and they get their own run first. Everything done and ahead: status and roadmap.
The site, https://inhousecloud.estoca.com.br/, has six parts, all from docs/:
Preview it locally: pip install -r docs/requirements.txt && mkdocs serve.
MIT, see LICENSE.