Repo stats from the GitHub API, cached Oct 2026.Homepage
About this project
Open Source AI Gateway for 100+ LLMs. Self-hosted. Enterprise-ready. Call any LLM in OpenAI format. LiteLLM Proxy Server (AI Gateway) | Hosted Proxy | Enterprise Tier | Website. LiteLLM is an open source AI Gateway that gives you a single, unified interface to call 100+ LLM providers — OpenAI, Anthropic, Gemini, Bedrock, Azure, and more — using the OpenAI format. Use it as a Python SDK for direct library integration, or deploy the AI Gateway (Proxy Server) as a centralized service for your team or organization.
litellm/router_strategy/complexity_router/jev_classifier.py routes requests using Jev judgments.
litellm/types/proxy/guardrails/guardrail_hooks/typesafe.py wires TypeSafe guardrail hooks on the proxy.
Dashboard components JevClassifierConfig.tsx expose classifier settings in the LiteLLM UI.
Watch out
Large gateway project; Jev is one optional path among many providers and guardrails. License is NOASSERTION in GitHub metadata.
Questions about this entry
Short answers drawn from the linked project or page.
Plain LiteLLM with Jev as one optional path. The README fetched here never mentions Jev at all; the integration lives in specific files, including a complexity-router Jev classifier and TypeSafe guardrail hooks on the proxy, beside 100+ other providers.
It supplies typed judgments that feed the complexity-based routing path, so the model used per request can depend on a Jev verdict. Which complexity buckets or thresholds it applies is not documented in the fetched README, and other router strategies are available instead.
Only for the Jev paths, which call TypeSafe and so need TypeSafe or gateway credentials alongside your provider keys. The rest of LiteLLM is self-hostable and its other routes, guardrails and routing strategies need no TypeSafe setup.
Both roles exist as separate pieces. Guardrail hooks on the proxy can reject traffic, while the Jev classifier's documented job is routing. Which Jev-specific policies actually block a request is left to your configuration, not fixed by the hook code.
The integration lives in litellm/router_strategy/complexity_router/jev_classifier.py, where Jev supplies typed judgments that feed the complexity-based routing path, so the model per request can depend on a Jev verdict.
A complexity-router Jev classifier, TypeSafe guardrail hooks wired by litellm/types/proxy/guardrails/guardrail_hooks/typesafe.py, and dashboard JevClassifierConfig.tsx components that expose classifier settings in the UI.
uv tool install 'litellm[proxy]' then litellm --model gpt-4o exposes an OpenAI-compatible endpoint on port 4000; the Jev paths sit alongside 100+ other providers and guardrails.
Reactions & coverage
Posts, threads, and videos about this entry from around the web.
X: Jev as an LLM guardrail inside Agentgateway
X: Jev lands on the Vercel AI Gateway
X: Agent Handoff Gate: agents verify what they hand off
X: Vercel's fx safety reviewer, measured at 18× faster
Alibaba's multi-agent platform, which implements native TypeSafe Jev classification through JevClassifierModel for binary, choice and score routing across agent pipelines. Python 3.11+, Apache-2.0, with a FastAPI agent service and web UI.
CLI-installed Gmail classifier that runs 24/7 on user-owned Google Apps Script: the worker asks Jev for three typed judgments per thread (requires_action, is_important, bucket), applies labels, stars, and optional archiving, and routes ambiguous or low-confidence mail to a Review label.
On-call tooling where Jev answers four typed questions per alert - actionable, severity, owning team and duplicate-of - and plain code routes, pages or drops. Includes a root-cause benchmark where Jev and code investigate recorded incidents with no general-purpose LLM in the loop, enforcing budgets and verification.
A Kubernetes admission webhook that asks Jev to allow, flag, or deny a manifest change for intent-shaped policies a Rego rule cannot express, keeping hard rules in Rego and failing closed by default.
Autonomous agent platform with 187,504 stars for building and running agents from plain English or a visual builder. Its platform features first-class TypeSafe Jev decision blocks for typed routing, filtering, scoring and confidence-gated next-action dispatching.
