sebabottero/ai-setup

Make your AI coding agent ask before git push, rm -rf or reading your SSH keys, and stop burning tokens without noticing.
One folder configures Claude Code, OpenCode and Codex the same way, on every machine.
| | What you get |
| 🛡 Asks before anything hard to undo | One policy.json and one guard decide allow / ask / deny for every shell command. Secrets are a hard deny. Claude also gets an OS sandbox from the same policy. |
| 💸 Spends fewer tokens | coach picks model and effort before you start. A hook warns when the context passes 100k. usagereport.py audits your own sessions, locally. |
| 🔀 Same rules in every tool | Write rules, skills and agents once. They install into Claude Code, OpenCode and Codex. Your private bits live in layers, not in a fork. |
| 🎫 One writer per ticket | flow-ticket plans, a single sub-builder writes in its own git worktree, and review, tests and checks run read-only in parallel. |
## See it work
The guard on real commands:
console
$ echo '{"tool":"bash","command":"git push origin main"}' | python3 core/guard/guard.py
{"decision": "ask", "reason": "🔴 high risk: 'git push origin main' needs your explicit OK (policy: git push )"}
$ echo '{"tool":"bash","command":"git status"}' | python3 core/guard/guard.py
{"decision": "allow", "reason": ""}
The token audit on a month of real use (python3 -I adapters/claude/usagereport.py 30; reads only your local transcripts, sends nothing):
console
Sessions: 149 | over 150k: 35 (23%) | over 300k: 9 | peak: 772k
Turns by model: claude-sonnet-5-5 60%, claude-opus-5-5 39%
Guard: 1233 allowed, 352 asked, 4 denied. Rules that fire most: outside-sandbox 123, git push 80
/coach before a task (example): model and effort, plus your request rewritten as goal · done when · out of scope · start from.
Every ~30 days tune-setup re-tunes it from your own /insights data, and edits only after you approve.
## Install
macOS with Homebrew and python3, or Linux with python3.
bash
git clone https://github.com/sebabottero/ai-setup.git ~/Dev/ai-setup # clone into the final path: the links point to it
cd ~/Dev/ai-setup
./bootstrap.py --dry-run # shows what it would install, changes nothing
./bootstrap.py # tools, then ./install.sh (runs the tests, backs up whatever it replaces)
## Is it for you?
Yes if you use an agent on real repos and want pushes, deletes and secret reads to need your OK, switch between agent tools,
work from tickets, or want to see and cut your token spend.