User repository on GitHub
Route Oh My Pi prompts between simple and advanced models with TypeSafe AI's Jev classifier.
- Forks
- 0
- Last pushed
- Updated Sep 2026
Repo stats from the GitHub API, cached Oct 2026.
CommunityJev Integrations & Adapters
Omp Plugin Jev Router
Oh My Pi extension that asks Jev whether each interactive prompt is simple or advanced, then switches the OMP model role and thinking level without rewriting the prompt.
- Stars
- 1
- Language
- TypeScript
- Published by
- Community
- Author
- atulify
- Last verified
- Oct 2026
About this project
Model routing for Oh My Pi, powered by TypeSafe AI's Jev classifier. The extension classifies each eligible prompt as simple or advanced, then selects the corresponding OMP model role before the prompt proceeds. Classification failures, ambiguous results, image prompts, and low-confidence simple results prefer the advanced route. Closed-set simple / advanced routing; Jev cannot choose arbitrary model IDs. Configurable confidence threshold, timeout, model specs, and thinking levels.
Quoted from the project's README. Read the full README.
Used inJev for Software & DevTools
- Languages
- TypeScript
Highlights
- Closed-set simple/advanced routing; failures, images, and low-confidence simple results prefer advanced.
- Defaults to @smol for simple and @advanced then @slow for advanced, both at high thinking level.
- Single Jev attempt with no retries, 0.75 default threshold, 2000ms timeout, and branch-local modes.
- Bounded context only and macOS Keychain support so the key stays out of shell configuration.
Watch out
Needs Oh My Pi 18.1.6, Node 20+, and configured simple/advanced model roles; auto-routing covers interactive input only. Copy reviewed against the repository README on 2026-10-04, not from running it.
Questions about this entry
Short answers drawn from the linked project or page.
No. Routing is fail-open: errors leave the prompt untouched and prefer the advanced route, falling from @advanced to @slow, then keeping the current model with a warning.
No. It only chooses between the configured simple and advanced roles, defaulting to @smol versus @advanced then @slow. It cannot invent model IDs.
Only the submitted text capped at 16,000 characters plus limited recent user and assistant context, with has_images false. Images, tool calls, and full transcripts stay local.
Keychain is macOS-only and recommended there. Elsewhere set TYPESAFE_API_KEY, and note OMP extensions run as trusted unsandboxed code in the OMP process.
Six: status shows the mode, credential source, model mappings, threshold and the last sanitized decision; auto and on classify each eligible prompt with Jev; off leaves the current model unchanged; simple and advanced pin eligible prompts to one route without calling Jev at all. Mode state is branch-local, and prompt text and credentials are never written into it.
Only interactive OMP input, in this version. Automatic routing applies to OMP 18.1.6 input events; that release does not emit the same event for print-mode or RPC prompts, so those paths are not routed automatically and need the manual commands instead.
Less than it looks, and the README is explicit. OMP extensions run as trusted, unsandboxed code inside the OMP process, so Keychain storage protects the credential at rest and avoids configuration or environment leakage but cannot hide it from the host process or another malicious extension there; strong isolation needs a separate credential-holding proxy. The extension reads the item with /usr/bin/security, keeps it in process memory and never copies it into process.env.
More Jev resources like this
46.7kGitHub stars
Privacy-first, self-hosted personal knowledge management system with block-level references and a local-first Markdown editor; the project integrates Jev's decision model for high-speed document classification, flashcard intent categorization and automated tag routing.
Jev Integrations & Adapters#community#note-taking#knowledge-base
Community
siyuan-note/siyuanTypeScript
siyuan-note/siyuanTypeScript19.6kGitHub stars
Open-source TypeScript agentic operating system: a core runtime, app for web and mobile, CLI, cloud services and first-party plugins. The directory's curated summary reports it integrates TypeSafe System One decision services for sub-100ms intent classification, action dispatching and confidence-gated tool execution.
Jev Integrations & Adapters#community#agents#framework
Community
elizaOS/elizaTypeScript
elizaOS/elizaTypeScript12.4kGitHub stars
Open-source AI email assistant (Next.js, Postgres, 12k+ stars) with a decision-model layer that can run on openrouter:typesafe/jev-1.13: Jev-backed questions cover rule selection, cold-email detection, thread status, recurring patterns, unsubscribe pages, and sender categorization, each with evals.
Jev Integrations & Adapters#community#email#gmail
Community
elie222/inbox-zeroTypeScript
elie222/inbox-zeroTypeScript689GitHub stars
The Vercel AI SDK provider for OpenRouter, including openrouter.decisionModel for typesafe/jev-1.13 evaluations with providerMetadata answers.
Jev Integrations & Adapters#community#typescript#openrouter
38GitHub stars
File-first project notebook where many agents share context in .context/ files while Jev only judges the next step and records it as JSON.
Jev Integrations & Adapters#community#jevbest#catalog-import
Community
pedrocivita/tocketTypeScript
pedrocivita/tocketTypeScript15GitHub stars
Merged PR in the Kilo Code cloud monorepo that replaces its LLM auto-routing classifier with one OpenRouter /systemone Choice request to typesafe/jev-1.13; the PR reports 0.93-0.94 route accuracy on 72 golden cases at roughly $0.00028 per call, with a 250 ms transport-error retry.
Jev Integrations & Adapters#community#routing#coding-agents
Community
Kilo-Org/cloudTypeScript
Kilo-Org/cloudTypeScriptFrom the Jev community
Posts from builders shipping with Jev right now.