# New Machine Setup

This guide is the shortest path for rolling the Codex Ops Kit onto another machine.

## 1. Clone The Repo

```bash
git clone https://github.com/slaguru666/codex-ops-kit.git
cd codex-ops-kit
```

## 2. Install The Kit

```bash
bash scripts/install_codex_kit.sh
```

What this does:

- backs up any existing `~/.codex/config.toml`
- backs up any existing `~/.codex/AGENTS.md`
- installs the managed profile block
- installs the global `AGENTS.md`
- copies the portable memory notes into `~/.codex/memories/portable-kit/`

## 3. Start Using Profiles

- `balanced`: default for most work
- `economy`: use for smaller tasks and cheaper runs
- `deep`: use for hard debugging, architecture, and more demanding tasks

## 4. Carry Forward Durable Memory

When you learn something worth reusing, update the markdown files in:

```bash
~/.codex/memories/portable-kit/
```

Then commit those changes back to this repo so your other machines can pull them.

## 5. Keep Local Storage Healthy

Audit the Codex home:

```bash
bash scripts/audit_codex_home.sh
```

Preview large old transcripts:

```bash
bash scripts/prune_codex_sessions.sh
```

Archive them after review:

```bash
bash scripts/prune_codex_sessions.sh --apply
```

## 6. Sync To Another Machine

On the machine that already has the latest memory updates:

```bash
git add memory docs README.md templates scripts
git commit -m "Update Codex kit memory and setup"
git push
```

On the next machine:

```bash
cd codex-ops-kit
git pull
bash scripts/install_codex_kit.sh
```

## Notes

- Do not sync `~/.codex/auth.json`.
- Do not sync live SQLite databases or raw `sessions/` history.
- Keep the portable layer small and curated so Codex does not waste tokens reloading unnecessary context.