auto-mode-for-paseo
Experimental Paseo plugin where Jev, local Laya or OpenAI Decisions classifies each message and routes it to a provider/model preset.
Overview
Auto Mode for Paseo is a plugin for Paseo, a multi-provider agent workspace. It reads each new message and sends it to the best preset for the task, where each preset bundles a provider (Codex, Claude, OpenCode or another installed one), a model, a reasoning setting and instructions. The conversation picker shows Auto and preset names, while model names stay in plugin settings.
A System 1 model only classifies the message and selects a preset: TypeSafe Jev by default, a local Laya model (no separate Python needed on Windows), or the experimental OpenAI Decisions API. The README is explicit that this is an experiment: the classifier does not read your code or plan the work, can pick the wrong preset, and should not be used in production; users should check each routing notice and choose manually when it matters.
What the decision model decides
| Type | Question | Answers |
|---|---|---|
| Choice | Which preset should handle this message? | user-defined presets |
How it works
- User sends a message in Auto mode
- Classifier (Jev, Laya or OpenAI Decisions) picks a preset
- Paseo runs the turn with that preset's provider and model
- A routing notice shows the choice
Strengths
- Swappable classifier, including fully local Laya
- Model details hidden behind presets
Caveats
- Explicitly experimental and not for production
- Classifier can choose the wrong preset
Getting started
npm ci && npm run check && paseo plugin install . (Paseo 0.8.0+, Node.js 24)At a glance
- Type
- Project
- Status
- prototype
- Language
- TypeScript
- Runtime
- Paseo plugin
- Has code
- yes
Use cases and integrations
How curated lists describe it
A Paseo provider that uses TypeSafe Jev to route each Codex turn.
hellogumbo/awesome-jev
Frequently asked questions
What does the classifier see?
Each new message; it does not read your code or plan the work.
Can I avoid a TypeSafe key?
Yes, use the local Laya classifier.
Is it production-ready?
No, the README says not to use it in production.
Related projects
Desktop workspace for coding agents that runs tasks in parallel worktrees, listed as using Jev to pick the best agent per task.
ProjectbetaTypeScript, RustCost-, cache- and quota-aware LLM router that classifies each turn with a Jev-class model, prices every route, and checks cheap answers.
ProjectprototypePython- jev-router★ 593
Per-turn model routing for Claude Code and Codex: wraps the real CLI and lets Jev send simple turns to fast models and hard ones to strong.
ProxybetaJavaScript Glue that turns one Linux machine into a team of coding agents, with Jev making typed calls like who builds a task and whether it may merge.
ProjectbetaTypeScriptOpen-source team agent platform for Slack, GitHub and more that asks Jev once per session which agent runtime and model to use.
ServicebetaTypeScriptOrchestrator that probes which coding agents are installed and logged in, then has Jev pick the best one for a task.
Agent skillbetaPython