Skip to content
Metatext Decide — typed decisions for your agents, with a probability on every answerGet early access
Metatext
Sign Up

x-herald

Chinese-documented LLM gateway with virtual model routing, protocol conversion and a transparent Jev System One proxy.

Servicebeta
View on GitHub ↗

Overview

x-herald is a transparent LLM gateway, documented in Chinese, that decouples public virtual model names from actual backend instances through a routing rule engine. It converts between OpenAI and Anthropic protocols, supports streaming, and includes a transparent JEV System One proxy at /api/v1/systemone that routes only to system_one model groups and substitutes the real Jev model name.

Around that it offers virtual keys with rate limits, retries with backoff, instance failover, health checks, circuit breakers, timeout budgets, request logs, performance baselines with anomaly alerts, cost tracking by key, provider and model, and a React admin UI. It runs on Bun with Hono and PostgreSQL or embedded PGlite, with unit and end-to-end tests. Jev here is proxied rather than used to make routing decisions.

What the decision model decides

TypeQuestionAnswers
OtherNone made by the gateway; it proxies Jev System One requests to configured instances.(inferred)—

How it works

  1. Configure a TypeSafe provider and system_one model group.
  2. Clients call /api/v1/systemone with a virtual key.
  3. The gateway selects an instance and forwards state and questions.
  4. Logs, costs and health are tracked.

Strengths

  • Centralized keys, rate limits and failover for Jev traffic.
  • Cost tracking and monitoring.

Caveats

  • Jev is proxied, not used for routing decisions.
  • Chinese documentation.

At a glance

Type
Service
Status
beta
Language
TypeScript
Runtime
Bun + Hono gateway, React admin
Has code
yes
Has tests
yes

Use cases and integrations

How curated lists describe it

Frequently asked questions

Does x-herald use Jev to route?

No, it proxies Jev System One requests like other providers.

What does it add for Jev traffic?

Virtual keys, rate limits, failover, logging and cost tracking.

What stack does it use?

Bun, Hono, PostgreSQL or PGlite, and React.

Related projects

  • ai

    Hack Club's lightweight LLM proxy, listed for adding a Jev forwarding endpoint behind its auth, limits and usage logging.

    MCP serverproductionTypeScript
  • decision-gate

    npm library that rate-limits, spend-caps, caches and redacts every Jev request made in a loop.

    LibrarybetaJavaScript
  • jev-accounts-hub

    Go gateway that pools multiple TypeSafe accounts behind the native System One API with user keys, quotas and failover.

    ServicebetaGo
  • Jeview

    Zero-dependency local gateway and live visualiser that proxies Jev calls, logs every decision to SQLite and draws them on a real-time map.

    ProxybetaJavaScript
  • jevkit

    Rust CLI for Jev that lints typed question sets offline, then sends them and prints probabilities, labels and scores as JSON.

    CLIbetaRust
  • jevtok-ts

    Offline TypeScript token counter and request-cost estimator for Jev 1.13, a port of the reconstructed jevtok encoder.

    PluginbetaTypeScript