Jev + Claude Code in seven minutes: enforcing routing with a hook
The author gives a seven-step setup that hooks Jev's Choice decisions into Claude Code's PreToolUse hook, so web fetches, subagents and retries must ask Jev first and are skipped if Jev says no.
Jev + Claude Code is the best agent setup I've built this year It just made my Claude Code runs CHEAPER and FASTER than what 95% of people are running... setup takes literally 7 minutes: prompt → Claude Code → Jev decision → Claude Code execution → result step 1 → open @typesafeai, create API key (never paste it into the chat) step 2 → export TYPESAFE_API_KEY in your shell, not in CLAUDE.md step 3 → prompt Claude Code: install typesafe-sdk + smoke test system_one (Choice) step 4 → tell Claude Code: build the usage lab (router, dry-run, config, logs) - or clone the GitHub below step 5 → add a PreToolUse hook + skill jev-usage-router: before web fetch / subagent / retry / heavy bash → call the router, honor the action the hook is the real trick: a skill is a suggestion, a hook is enforced. if Jev says no, the tool call doesn't happen step 6 → run shadow first, read the logs, go active when you trust it - kill switch: remove the hook or enabled: false step 7 → flip active: Claude Code asks - Jev decides - Claude Code executes - you approve anything irreversible the result: Jev + Claude Code running right in your terminal, I've already tested it on routine tasks - and the results are genuinely incredible You can come up with endless ways to use Jev + Claude Code - but the most important thing is to set it up as soon as possible Copy this setup, explore my repo below - then read the full Jev deep dive ↓
This post quotes another post: x.com/asteri_eth/status/2104901781725598103