DEMO / RECORDING
Overview & result
A developer tool mod for Claude Code that classifies prompt complexity in 35ms, sending lightweight commands to cheap fast models and reserving heavy reasoning for deep refactoring.
From the original post
Introducing Jev Model Router for Claude Code
This Claude Code Mod lets you use Jev through its direct @typesafeai API or @vercel AI Gateway
With every request you send to Claude Code, Jev classifies the subagent model, main model (only at session start to avoid breaking the cache), and effort level
Install it with one command:
npx claude-code-templates@latest --mod productivity/jev-model-router
Full Mod details:
How Jev fits in the loop
- Intercept developer instructions entered in Claude Code CLI.
- Classify required effort tier and model fit via Jev Choice.
- Dispatch request through TypeSafe API or Vercel AI Gateway.
- Execute the coding action with zero perceived user lag.
How to reproduce
- Inspect terminal recording showing live prompt interception in Claude Code.
- Configure TYPESAFE_API_KEY or Vercel AI Gateway endpoint in shell profile.
- Verify simple file edits route instantly to fast tiers without multi-second pauses.
Why this build matters
Solves developer friction in autonomous coding by removing unnecessary 30-second reasoning pauses on trivial one-line edits.
Limitations
- Requires Claude Code CLI hook support.
- Complex prompts with ambiguous intent may fall back to default reasoning tier.
- Author reported demo; dependent on network latency to TypeSafe API.