Web / Done

MyRouter AI Routing Gateway

Role | Full-Stack Developer Timeline | 2026
MyRouter AI Routing Gateway

Overview

Self-hosted AI routing gateway: one OpenAI-compatible endpoint for many upstream providers (OpenAI, Anthropic, Gemini, DeepSeek, any OpenAI-compatible gateway) with automatic fallback, per-key access limits, and a dashboard. Ships as a single Go binary with the React console embedded and SQLite storage, so there are no external services to operate.

Challenge

Giving every client one stable base URL while upstreams fail in different ways: rate limits, timeouts, 5xx, network errors, and billing cutoffs each need their own fallback trigger, per-key and per-route guard policies must hold under concurrency, and cost accounting has to stay honest against official catalog prices instead of inventing numbers.

Result

Delivered a working gateway with virtual model aliases backed by ordered primary to fallback chains, server-side skills injected as system context on every request, live observability with per-request traces, and reproducible demo screenshots generated through the public admin API by seed scripts rather than mocked UI state.

Key Statistics

OpenAI, Anthropic, Gemini wire formats

Upstream protocols

v0.1.10

Release

Embedded SQLite, no external services

Storage

go test plus npm typecheck

Verification

Technologies

Backend

  • Go 1.23+
  • SQLite

Frontend

  • React 18
  • TypeScript 5.6
  • Tailwind CSS 3.4

AI / Gateway

  • OpenAI-compatible endpoint
  • Fallback chains
  • Server-side skills
  • models.dev catalog pricing

Quality

  • go test
  • npm typecheck
  • E2E API harness
  • Secret scan scripts

Deployment

  • Single binary
  • Docker
  • Windows tray app

Gallery

Dashboard

Dashboard

Live routing, provider health, throughput chart with 24h to 30d ranges, and recent activity.

Providers

Providers

Upstream endpoints, credentials, health, and per-provider tabs for overview, models, and guard policy.

Routes

Routes

Virtual model aliases with ordered primary to fallback chains and per-category triggers.

Guard

Guard

RPM and TPM caps, daily and monthly token limits, concurrency, IP and device limits with auto-blocking.

Skills

Skills

Built-in and custom skills with bindings plus a knowledge drawer for stored items.

Test Lab

Test Lab

Playground for sending real prompts through a route and seeing which provider and model answered.

More projects

Back to all projects