lintasan-go

README

🇮🇩 Lintasan

Setiap Koneksi Punya Jalannya — Jalur cerdas yang menghubungkan manusia, AI, dan sistem dalam satu aliran terintegrasi.

⚡ 35× lebih ringan dari Node.js v1 · single binary ~24MB · 900+ tests · 118+ provider presets · Go v2 + SvelteKit Embedded SPA

---

🇬🇧 Lintasan

Every Connection Has Its Path — An intelligent pathway connecting humans, AI, and systems in one integrated flow.

⚡ 35× lighter than Node.js v1 · single ~24MB binary · 900+ tests · 118+ provider presets · Go v2 + SvelteKit Embedded SPA

---

🌐 Filosofi / Philosophy

🇮🇩 Bahasa Indonesia

"Lintasan adalah jalur tempat koneksi, kecerdasan, dan kemungkinan bergerak."

Di dunia modern, AI bukan hanya tentang model. AI adalah tentang bagaimana sistem saling terhubung, bagaimana request menemukan jalur terbaik, dan bagaimana manusia berinteraksi dengan teknologi secara efisien.

Lintasan hadir sebagai jalur cerdas yang menghubungkan manusia, AI, dan sistem dalam satu aliran yang terintegrasi. Satu endpoint untuk semua provider AI — tidak perlu lagi berganti-ganti SDK atau mengelola banyak API key di berbagai tempat.

🇬🇧 English

"Lintasan is the path where connections, intelligence, and possibilities move."

In the modern world, AI isn't just about models. It's about how systems connect, how requests find their best path, and how humans interact with technology efficiently.

Lintasan exists as an intelligent pathway connecting humans, AI, and systems in one integrated flow. One endpoint for all AI providers — no more switching SDKs or managing multiple API keys across different places.

---

📖 Daftar Isi / Table of Contents

• Apa itu Lintasan? / What is Lintasan?
• Kenapa Go? / Why Go?
• Tech Stack
• Fitur Utama / Key Features
• Arsitektur / Architecture
• Quick Start
• 5-Minute Quickstart
• Instalasi / Installation
• Konfigurasi / Configuration
• API Reference
• Dashboard
• Provider Presets
• Struktur Project / Project Structure
• Development
• Testing
• Deployment
• Benchmark
• Migration dari Node.js v1
• Contributing
• Lisensi / License

---

❓ Apa itu Lintasan? / What is Lintasan?

🇮🇩 Bahasa Indonesia

Lintasan adalah LLM proxy gateway dengan 40+ fitur optimasi. Satu endpoint OpenAI-compatible untuk semua provider AI — routing cerdas, embedded cache, dual-mode CommandCode, plugin system, dan penghematan token otomatis.

Masalah yang diselesaikan:

• 🔀 Multi-provider complexity — ganti provider = ganti SDK, ganti API key, ganti format
• 💸 No cost visibility — tidak tahu berapa token terpakai per model/provider
• 🔄 No failover — satu provider down, semua request gagal
• 🔐 Key management chaos — API key tersebar di mana-mana
• 📊 No observability — tidak bisa lihat usage, latency, error rate

Solusi Lintasan:

• Satu endpoint → semua provider (OpenAI, Anthropic, DeepSeek, Gemini, Groq, dll)
• Satu API key → autentikasi terpusat
• Smart routing → request otomatis ke provider terbaik
• Circuit breaker → provider gagal auto-disable
• Dashboard → monitoring real-time 17 halaman
• Plugin system → extensible tanpa ubah core

🇬🇧 English

Lintasan is an LLM proxy gateway with 40+ optimization features. One OpenAI-compatible endpoint for all AI providers — smart routing, embedded cache, dual-mode CommandCode, plugin system, and automatic token savings.

Problems solved:

• 🔀 Multi-provider complexity — switching providers means switching SDKs, API keys, and formats
• 💸 No cost visibility — can't see token usage per model/provider
• 🔄 No failover — one provider down, all requests fail
• 🔐 Key management chaos — API keys scattered everywhere
• 📊 No observability — can't see usage, latency, error rates

Lintasan's solution:

• One endpoint → all providers (OpenAI, Anthropic, DeepSeek, Gemini, Groq, etc.)
• One API key → centralized authentication
• Smart routing → requests automatically go to the best provider
• Circuit breaker → failing providers auto-disable
• Dashboard → 17-page real-time monitoring
• Plugin system → extensible without core changes

---

💪 Kenapa Go? / Why Go?

Metric	Node.js (legacy)	Go (current)	Improvement
RAM	~500MB	~14MB	35x lebih hemat
Binary size	513MB (node_modules)	24MB (single file)	21x lebih kecil
Startup	3-5 detik	<50ms	60-100x lebih cepat
Concurrent req/s	~10,000	~50,000+	5x throughput
Dependencies	800+ npm packages	1 (go-sqlite3)	800x lebih sedikit
Tests	Manual	700+ / 35 packages	Automated
Provider presets	27	118	4x lebih banyak
Deployment	Docker + npm install	scp 24MB binary	Zero setup

---

🛠 Tech Stack

Layer	Technology	Notes
Backend	Go 1.22+	HTTP server, routing, proxy, streaming
Database	SQLite (go-sqlite3)	Embedded, zero config, single-file
Frontend	SvelteKit 5 + TypeScript	Dashboard SPA, client-rendered, embedded into the Go binary via go:embed
Styling	Tailwind CSS v4 + CSS variables	Dark/light mode, responsive
CLI	Cobra	start, setup, mitm, version
Testing	Go standard library	581 tests, 39 packages
Deployment	Single self-contained binary + systemd	UI + API in one executable; no Node.js required. Pure proxy

---

✨ Fitur Utama / Key Features

⚠️ Kebijakan Legal & Etika / Legal & Ethics Policy

🇮🇩 Lintasan didesain untuk menggunakan API resmi yang sah (Legal API). Kami secara tegas tidak melakukan/mendukung Reverse Engineering terhadap endpoint internal IDE komersial. Lintasan berfokus pada integrasi provider API resmi dan Experimental Provider (ACP).

🇬🇧 Lintasan is designed to use legitimate, official APIs. We strictly do not support reverse engineering of commercial IDE internal endpoints. Lintasan focuses on official API provider integrations and Experimental Provider pipelines (ACP).

🇮🇩 Fitur

#	Fitur	Deskripsi
1	Multi-Provider Proxy	Satu endpoint untuk 10+ provider LLM
2	Smart Routing	Multi-stage: regex → header → model → fallback
3	113 Provider Presets	Semua provider LiteLLM siap pakai
4	Connection Management	Add/test/sync/delete connections + auto discovery
5	Model Discovery	Auto-fetch models dari provider /models endpoint
6	Provider Test	Real-time latency + model count testing
7	Streaming (SSE)	Full Server-Sent Events untuk streaming
8	Fallback Chains	Multi-level fallback per model
9	Circuit Breaker	Auto-disable provider yang gagal
10	Request Logging	Complete request/response logging
11	Analytics	Real-time metrics: latency, tokens, throughput
12	Combo System	Pre-configured model+provider bundles
13	Load Balancer	Model-aware weighted load balancing
14	Plugin System	Plugin store + auto-registration
15	Vector Memory	Pluggable embedder dengan SQLite default
16	Web Search	Augment chat dengan live web results
17	OAuth Integration	Provider OAuth flow support
18	Image Generation	Proxy ke DALL-E / Stable Diffusion
19	Audio (TTS + STT)	Speech + transcription via OpenAI API
20	Token Budgeting	Per-key daily/monthly limits
21	Cost Tracking	Real-time cost tracking per request
22	API Keys	Key management + usage tracking
23	Teams & Users	Multi-user access control
24	Webhooks	Event-driven webhook system
25	Backup & Export	Database backup + disaster recovery
26	Dashboard	17-page interactive SvelteKit dashboard
27	Playground	Built-in API test console
28	CORS	Built-in — use from any browser app
29	Zero Config	SQLite embedded — no setup required
30	CLI (Cobra)	start, setup, mitm, version
31	MITM Bridge	Optional HTTPS bridge untuk LocalAI/LM Studio
32	Experimental Mode	Cohort Provider ACP execution environment
33	Vector Search	In-memory TF-IDF + auto context injection
34	Credential Vault	AES-256-GCM UI Credential Dashboard
35	cURL Import	Quick-add connections from cURL commands
36	Metrics Endpoint	Exposes /metrics endpoint and Dashboard Observability UI

🇬🇧 Features

#	Feature	Description
1	Multi-Provider Proxy	One endpoint for 10+ LLM providers
2	Smart Routing	Multi-stage: regex → header → model → fallback
3	113 Provider Presets	All LiteLLM providers ready to use
4	Connection Management	Add/test/sync/delete connections + auto discovery
5	Model Discovery	Auto-fetch models from provider /models endpoint
6	Provider Test	Real-time latency + model count testing
7	Streaming (SSE)	Full Server-Sent Events for streaming
8	Fallback Chains	Multi-level fallback per model
9	Circuit Breaker	Auto-disable failing providers
10	Request Logging	Complete request/response logging
11	Analytics	Real-time metrics: latency, tokens, throughput
12	Combo System	Pre-configured model+provider bundles
13	Load Balancer	Model-aware weighted load balancing
14	Plugin System	Plugin store + auto-registration
15	Vector Memory	Pluggable embedder with SQLite default
16	Web Search	Augment chat with live web results
17	OAuth Integration	Provider OAuth flow support
18	Image Generation	Proxy to DALL-E / Stable Diffusion
19	Audio (TTS + STT)	Speech + transcription via OpenAI API
20	Token Budgeting	Per-key daily/monthly limits
21	Cost Tracking	Real-time cost tracking per request
22	API Keys	Key management + usage tracking
23	Teams & Users	Multi-user access control
24	Webhooks	Event-driven webhook system
25	Backup & Export	Database backup + disaster recovery
26	Dashboard	17-page interactive SvelteKit dashboard
27	Playground	Built-in API test console
28	CORS	Built-in — use from any browser app
29	Zero Config	SQLite embedded — no setup required
30	CLI (Cobra)	start, setup, mitm, version
31	MITM Bridge	Optional HTTPS bridge for LocalAI/LM Studio
32	Experimental Mode	Cohort Provider ACP execution environment
33	Vector Search	In-memory TF-IDF + auto context injection
34	Credential Vault	AES-256-GCM UI Credential Dashboard
35	cURL Import	Quick-add connections from cURL commands
36	Metrics Endpoint	Exposes /metrics endpoint and Dashboard Observability UI

---

🏗 Arsitektur / Architecture

Client (App / Agent / curl / IDE)
        │
    ▼             ▼
┌─────────────────────────────────────────────────────┐
│ Nginx (SSL Termination) — lintasan.sans.biz.id      │
│   * All traffic proxies to Go:20180                 │
└─────────────────────────┬───────────────────────────┘
                          │
         ┌────────────────▼──────────────────┐
         │ Go Backend :20180 (lintasan start)│
         │ ── Serves BOTH API & UI ──        │
         │ • Embedded SPA UI (go:embed)      │
         │ • OpenAI-compatible LLM proxy     │
         └──────────────┬────────────────────┘
     │
     │  ┌─────────────────────────┐
     │  │  API Gateway            │
     │  │  /v1/chat/completions   │
     │  │  /api/connections/import-curl │
     │  │  /v1/embeddings         │
     │  │  /v1/images/generations │
     │  │  /v1/audio/*            │
     │  │  /metrics               │
     │  │  /v1/models             │
     │  │  /v1/memory/*           │
     │  └──────────┬──────────────┘
     │             │
     │  ┌──────────▼──────────────┐
     │  │  Smart Router           │
     │  │  1. Header-based        │
     │  │  2. Model name match    │
     │  │  3. Load-balanced pick  │
     │  │  4. Priority sort       │
     │  │  5. Fallback chain      │
     │  └──────────┬──────────────┘
     │             │
     │  ┌──────────▼──────────────┐
     │  │  Optimization Pipeline  │
     │  │  • Circuit Breaker      │
     │  │  • Settings Cache (5s)  │
     │  │  • Request Logging      │
     │  │  • Cost Tracking        │
     │  │  • Vector Context Inject│
     │  │  • RTK Compressor       │
     │  └──────────┬──────────────┘
     │             │
     │  ┌──────────▼──────────────┐
     │  │  Provider Dispatcher    │
     │  │  + HTTP/1.1 keep-alive  │
     │  │  + Connection pooling   │
     │  │  + Request translation  │
     │  │  + SSE streaming        │
     │  └──────────┬──────────────┘
     │             │
     └─────────────┼───────────────────
                   │
         ┌─────────┼─────────┬──────────┐
         ▼         ▼         ▼          ▼
    ┌────────┐ ┌──────┐ ┌──────┐ ┌──────────┐
    │ OpenAI │ │Gemini│ │Groq  │ │Sumopod   │ ...118 providers
    └────────┘ └──────┘ └──────┘ └──────────┘

---

🚀 Quick Start

⏱ Prefer a step-by-step walkthrough? See QUICKSTART.md — from zero to first LLM request in <5 minutes.

The dashboard UI is embedded inside the binary, so one executable serves the full app (UI + API) on :20180 — no Node, no nginx.

# Download the pre-built binary and run
curl -L -o lintasan https://github.com/sanhaji182/lintasan/releases/latest/download/lintasan-linux-amd64
chmod +x lintasan
./lintasan start

# Dashboard → http://localhost:20180/dashboard
# API       → http://localhost:20180/v1/chat/completions

---

🎥 Setup demo coming soon — GIF showing Lintasan from download to first chat request.

---

📦 Instalasi / Installation

Full guide: INSTALL.md — pre-built binary, build from source, Docker, systemd, nginx.

3 methods — pick one:

🇮🇩 3 Cara Install

Opsi 1: Binary Pre-built (Recommended)

Satu binary, dashboard sudah termasuk di dalamnya.

curl -L -o lintasan https://github.com/sanhaji182/lintasan/releases/latest/download/lintasan-linux-amd64
chmod +x lintasan
./lintasan start

Opsi 2: Build dari Source

Butuh Go 1.22+ dan Node 20+ (untuk build dashboard). make build meng-compile frontend SvelteKit jadi SPA statis, meng-embed-nya ke binary Go, lalu compile binary tunggal.

git clone https://github.com/sanhaji182/lintasan.git
cd lintasan
make build       # frontend → embed → ./lintasan
./lintasan start

Tanpa Node? go build -o lintasan ./cmd/lintasan tetap jalan, tapi menghasilkan server API-only (tanpa UI dashboard).

Opsi 3: Docker (single container)

git clone https://github.com/sanhaji182/lintasan.git
cd lintasan
LINTASAN_MASTER_KEY=$(openssl rand -hex 32) docker compose up --build
# UI + API → http://localhost:20180

CLI Commands

./lintasan start      # Start server (UI + API, default :20180)
./lintasan setup      # Initialize database
./lintasan mitm start # Start MITM HTTPS bridge (optional)
./lintasan version    # Show version
./lintasan help       # All commands

# Custom port
PORT=8080 ./lintasan start

🇬🇧 3 Installation Methods

Option 1: Pre-built Binary (Recommended)

One binary, dashboard included.

curl -L -o lintasan https://github.com/sanhaji182/lintasan/releases/latest/download/lintasan-linux-amd64
chmod +x lintasan
./lintasan start

Option 2: Build from Source

Requires Go 1.22+ and Node 20+ (to build the dashboard). make build compiles the SvelteKit frontend into a static SPA, embeds it into the Go binary, and compiles a single executable.

git clone https://github.com/sanhaji182/lintasan.git
cd lintasan
make build       # frontend → embed → ./lintasan
./lintasan start

No Node? go build -o lintasan ./cmd/lintasan still works but produces an API-only server (no dashboard UI).

Option 3: Docker (single container)

git clone https://github.com/sanhaji182/lintasan.git
cd lintasan
LINTASAN_MASTER_KEY=$(openssl rand -hex 32) docker compose up --build
# UI + API → http://localhost:20180

CLI Commands

./lintasan start      # Start server (UI + API, default :20180)
./lintasan setup      # Initialize database
./lintasan mitm start # Start MITM HTTPS bridge (optional)
./lintasan version    # Show version
./lintasan help       # All commands

# Custom port
PORT=8080 ./lintasan start

---

⚙ Konfigurasi / Configuration

Full guide: CONFIGURATION.md — env vars, dashboard settings, connection setup, routing, security.

Quick env reference:

🇮🇩 Environment Variables

Variable	Default	Keterangan
PORT	20180	Port server utama
LINTASAN_DATA_DIR	./data	Direktori data (DB, logs)
LINTASAN_MASTER_KEY	auto-generated	Master API key
LINTASAN_OAUTH_IDE_ENABLED	false	Fallback jika setting dashboard belum disimpan; prefer Settings → Experimental → OAuth IDE
LINTASAN_OAUTH_PUBLIC_BASE_URL	—	Public origin for OAuth redirect when IDE OAuth is enabled
MITM_PORT	8443	MITM bridge port

Tidak perlu .env file — set env vars atau gunakan default. Database auto-create saat pertama run.

🇬🇧 Environment Variables

Variable	Default	Description
PORT	20180	Main server port
LINTASAN_DATA_DIR	./data	Data directory (DB, logs)
LINTASAN_MASTER_KEY	auto-generated	Master API key
LINTASAN_OAUTH_IDE_ENABLED	false	Fallback jika setting dashboard belum disimpan; prefer Settings → Experimental → OAuth IDE
LINTASAN_OAUTH_PUBLIC_BASE_URL	—	Public origin for OAuth redirect when IDE OAuth is enabled
MITM_PORT	8443	MITM bridge port

No .env file needed — just set env vars or use defaults. Database auto-creates on first run.

---

📡 API Usage

# Chat completion (OpenAI-compatible)
curl http://localhost:20180/v1/chat/completions \
  -H "Authorization: Bearer YOUR_MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{
    "model": "gpt-4o",
    "messages": [{"role": "user", "content": "Hello!"}],
    "stream": true
  }'

# List models
curl http://localhost:20180/v1/models \
  -H "Authorization: Bearer YOUR_MASTER_KEY"

# Embeddings
curl http://localhost:20180/v1/embeddings \
  -H "Authorization: Bearer YOUR_MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "text-embedding-3-small", "input": "Hello world"}'

# Image generation
curl http://localhost:20180/v1/images/generations \
  -H "Authorization: Bearer YOUR_MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "dall-e-3", "prompt": "A sunset over mountains"}'

# Text-to-speech
curl http://localhost:20180/v1/audio/speech \
  -H "Authorization: Bearer YOUR_MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "tts-1", "input": "Hello world", "voice": "alloy"}'

# Speech-to-text
curl http://localhost:20180/v1/audio/transcriptions \
  -H "Authorization: Bearer YOUR_MASTER_KEY" \
  -F "file=@audio.mp3" -F "model=whisper-1"

# Provider presets
curl http://localhost:20180/api/providers/presets

# Connection test
curl -X POST http://localhost:20180/api/connections/test \
  -H "Content-Type: application/json" \
  -d '{"base_url": "https://api.openai.com/v1", "api_key": "sk-..."}'

# Vector memory search
curl "http://localhost:20180/v1/memory/search?q=hello&limit=5" \
  -H "Authorization: Bearer ***
  
# Import curl via API Connection Creation
curl -X POST "http://localhost:20180/api/connections/import-curl" \
  -H "Authorization: Bearer ***

# Web search augmented chat
curl http://localhost:20180/v1/web/search \
  -H "Authorization: Bearer YOUR_MASTER_KEY" \
  -H "Content-Type: application/json" \
  -d '{"model": "gpt-4o", "query": "latest AI news"}'

🇮🇩 Drop-in replacement untuk OpenAI API — ganti base URL, semua SDK (openai-python, langchain, llama-index, dll) langsung kompatibel.

🇬🇧 Drop-in replacement for OpenAI API — change the base URL and all SDKs work immediately.

---

📊 Dashboard

🇮🇩 17 Halaman Dashboard

Lintasan dilengkapi dashboard interaktif berbasis SvelteKit 5 + Tailwind v4 untuk monitoring dan konfigurasi real-time.

|| Halaman | Fungsi | ||---------|--------| || Overview | Statistik global — requests, tokens, cache hit rate, latency | || Accounts | Kelola koneksi provider (add/test/sync/delete) | || Providers | 118+ provider presets, one-click add | || Experimental | Laboratorium fitur eksperimental | || Discover | Auto-discover free & public provider API | || OAuth IDE | OAuth flow untuk IDE agents (LAB) | || Routing | Konfigurasi combo + load balancer + aliases | || Fallback | Multi-level fallback chain per model/connection | || Logs | Real-time request log dengan filter & search | || Usage | Token usage + cost per provider/model | || Observability | Exportable /metrics + Real-time Grafana-like panels | || Analytics | Metrics dashboard — latency, throughput, savings | || Memory | Vector memory untuk RAG (SQLite/Redis) | || API Keys | Generate, copy, revoke API keys | || Teams | Team-based access control | || Users | User management + role assignment | || Webhooks | Event-driven webhook setup & testing | || Backup | Database backup, restore, export | || Settings | Global configuration — port, keys, limits | || Plugins | Plugin store + management + AI generator | || Playground | Interactive chat console untuk testing API | || MCP Server | Model Context Protocol server (14+ tools) | || Cost Savings | Breakdown penghematan token & cache | || Format Translator | Terjemahan format API lintas provider | || Docs | Dokumentasi API built-in |

Akses: http://localhost:20180/dashboard — embedded di binary Go (tanpa Node.js terpisah).

🇬🇧 17 Dashboard Pages

Lintasan comes with an interactive dashboard built with SvelteKit 5 + Tailwind v4 for real-time monitoring and configuration.

|| Page | Function | ||------|----------| || Overview | Global stats — requests, tokens, cache hit rate, latency | || Accounts | Manage provider connections (add/test/sync/delete) | || Providers | 118+ provider presets, one-click add | || Experimental | Experimental features lab | || Discover | Auto-discover free & public provider APIs | || OAuth IDE | OAuth flow for IDE agents (LAB) | || Routing | Smart-route configs: combo, load balancer, aliases | || Fallback | Multi-level fallback chains per model/connection | || Logs | Real-time request log with filter & search | || Usage | Token usage & cost per provider/model | || Observability | Exportable /metrics + real-time Grafana-like panels | || Analytics | Metrics dashboard — latency, throughput, savings | || Memory | Vector memory for RAG (SQLite/Redis) | || API Keys | Generate, copy, revoke API keys | || Teams | Team-based access control | || Users | User management + role assignment | || Webhooks | Event-driven webhook setup & testing | || Backup | Database backup, restore, export | || Settings | Global configuration — port, keys, limits | || Plugins | Plugin store + management + AI generator | || Playground | Interactive chat console for API testing | || MCP Server | Model Context Protocol server (14+ tools) | || Cost Savings | Token compression & cache savings breakdown | || Format Translator | Cross-format API translation tool | || Docs | Built-in API documentation |

Access: http://localhost:20180/dashboard — embedded in the Go binary (no separate Node process).

---

🔌 Provider Presets (118 Ready)

🇮🇩 Daftar Provider

Major Providers (4)

OpenAI · Anthropic · DeepSeek · Google Gemini

Top-Tier (6)

xAI (Grok) · Mistral AI · Azure OpenAI · Azure AI Foundry · Google Vertex AI · AWS Bedrock

AI Coding (4)

Codestral API · GitHub Copilot API · Pydantic AI Agents · Meta Llama API

Aggregators (8)

OpenRouter · Replicate · HuggingFace Inference · Vercel AI Gateway · AIML API · Poe by Quora · CometAPI · NanoGPT

High-Speed Inference (11)

Groq · Together AI · Fireworks AI · Cerebras · NVIDIA NIM · Cloudflare Workers AI · Hyperbolic · Lambda AI · FriendliAI · Anyscale Endpoints · SambaNova

GPU Cloud (12)

Baseten · OctoAI · Lepton AI · Featherless AI · Crusoe Cloud · nscale AI · PublicAI · Galadriel · Chutes · GMI Cloud · Heroku AI · Novita AI

CommandCode (2)

CommandCode (API Key) · CommandCode (Alpha)

Chinese Providers (10)

GLM / Zhipu AI · Kimi / Moonshot · MiniMax · Qwen / Alibaba · SiliconFlow · Xiaomi MiMo · Volcano Engine (ByteDance) · Z.AI · DeepSeek · Baidu Qianfan

Indonesia Providers (2)

Sumopod · Apertis AI (Stima API)

Enterprise & Cloud (8)

Snowflake Cortex AI · Oracle Cloud (OCI) · SAP AI Core · IBM watsonx · Gradient AI · NLP Cloud · Petals · Clarifai

Specialized (19)

Perplexity · Cohere · DeepInfra · Nebius AI · Aleph Alpha · AI21 Labs · Reka AI · Voyage AI · Deepgram · Black Forest Labs · Stability AI · Runway ML · Recraft AI · fal.ai · Helicone · Lemonade AI · Bytez · Sarvam AI · MorphDB

Self-Hosted (3)

Ollama · vLLM · LM Studio

Other (24)

AWS SageMaker · GigaChat · Predibase · OpenPipe · Scale AI · Titan ML · OctoML · Monster API · GooseAI · Forefront · Custom

---

📂 Struktur Project / Project Structure

lintasan/
├── cmd/lintasan/              # CLI entry point (Cobra)
├── internal/
│   ├── auth/                  # OAuth + API key auth
│   ├── config/                # Environment & DB config
│   ├── dashboard/             # (deprecated, moved to frontend/)
│   ├── db/                    # SQLite database layer
│   ├── discover/              # Model auto-discovery
│   ├── freeproviders/         # Free provider scanner (118+ presets)
│   ├── expprovider/           # AES-256-GCM Credentials
│   ├── memory/                # Vector memory (pluggable embedder)
│   ├── mitm/                  # MITM HTTPS bridge
│   ├── mlrouter/              # ML-based routing
│   ├── plugin/                # JS plugin engine
│   ├── rtk/                   # RTK token compressor
│   ├── server/                # HTTP server + all handlers
│   │   ├── server.go          # Core server + middleware
│   │   ├── proxy.go           # Provider proxy + streaming
│   │   ├── router.go          # Smart routing logic
│   │   ├── cache.go           # Settings cache
│   │   ├── handlers_*.go      # API handlers (dashboard, parity, etc.)
│   │   └── middleware_*.go    # CORS, auth middleware
│   └── websearch/             # Web search engine
├── frontend/                  # SvelteKit dashboard (Go+Svelte stack)
│   ├── src/
│   │   ├── routes/            # 17 page routes
│   │   ├── lib/               # Components, stores, utils
│   │   └── app.html           # Root HTML template
│   ├── static/                # Static assets
│   ├── svelte.config.js       # SvelteKit config (adapter-node)
│   ├── vite.config.ts         # Vite config
│   └── package.json           # Frontend dependencies
├── data/                      # Runtime data DB (auto-created)
├── docs/                      # Documentation
│   ├── api-reference.md       # Full API reference (deprecated, read AGENTS.md)
│   └── design-system.md       # UI design system
├── .env.example               # Example environment file
├── Dockerfile                 # Docker build
├── docker-compose.yml         # Docker compose
├── go.mod / go.sum            # Go dependencies
├── Makefile                   # Build automation
├── LICENSE                    # MIT
└── README.md                  # This file

---

💻 Development

# Clone
git clone https://github.com/sanhaji182/lintasan.git
cd lintasan

# Backend
go run ./cmd/lintasan start        # Run server (hot-reload with air if installed)

# Frontend
cd frontend
npm run dev -- --port 5173        # Dev UI + Vite server with HMR (Requires GO RUNNING)

# Build all
make build                         # go build + npm run build

# Run tests
go test ./...                      # 700+ tests
cd frontend && npm run check       # SvelteKit type-check

---

🧪 Testing

# All backend tests
go test ./...                      # 700+ passed, 0 failed, 39 packages

# With coverage
go test -cover ./...

# Specific package
go test -v ./internal/server/...

# Frontend
cd frontend && npm run check       # SvelteKit type-check + lint

---

🚢 Deployment

🇮🇩 Production Deployment

# Build statically linked production binary
cd lintasan
go build -ldflags="-s -w" -o lintasan ./cmd/lintasan

# Build frontend inside go executable
cd frontend && npm install && npm run build && cd ..

# Copy to server
scp lintasan user@server:/opt/lintasan/
scp -r frontend/build user@server:/opt/lintasan/frontend-build/

# Systemd service
sudo tee /etc/systemd/system/lintasan.service << 'EOF'
[Unit]
Description=Lintasan LLM Proxy Gateway
Documentation=https://github.com/sanhaji182/lintasan
After=network.target

[Service]
Type=simple
User=lintasan
WorkingDirectory=/opt/lintasan
ExecStart=/opt/lintasan/lintasan start
Restart=on-failure
RestartSec=5

[Install]
WantedBy=multi-user.target
EOF

sudo systemctl enable --now lintasan

Nginx Reverse Proxy

server {
    listen 443 ssl;
    server_name your-domain.com;

    # API → Go backend
    location /api/ { proxy_pass http://127.0.0.1:20180; }
    location /v1/  { proxy_pass http://127.0.0.1:20180; }

---

🏆 Benchmark

Duel head-to-head melawan 9Router — backend sama (CommandCode + deepseek-v4-pro).

Head-to-head duel against 9Router — same backend (CommandCode + deepseek-v4-pro).

Task	Lintasan	9Router	Winner
LRU Cache implementation	11.2s	11.6s	Lintasan
Code review merge_sorted	7.2s	10.9s	Lintasan
TypeScript deep-merge generic	14.7s	15.4s	Lintasan
Optimistic vs pessimistic locking	11.1s	15.3s	Lintasan
Docker exits code 0	12.9s	6.3s	9Router
GitHub Actions workflow	8.0s	3.7s	9Router
Rate limiting middleware	6.0s	4.1s	9Router
REST vs GraphQL vs gRPC	11.5s	15.4s	Lintasan

Results: Cold avg 10.3s vs 10.3s (parity) · Lintasan 5 — 9Router 3 · Thinking leaked: Lintasan 1/8 vs 9Router 3/8

---

🔄 Migration dari Node.js v1

🇮🇩 Panduan Migrasi

Lintasan Go menggunakan database schema yang berbeda dari Node.js v1 — tidak backward-compatible.

1. Export settings dari Node v1 dashboard
2. Install Lintasan Go (lihat Instalasi)
3. Re-create connections di Go v2 dashboard atau via API
4. Sync models — auto-discovered saat pertama connect
5. Verifikasi — test lewat Playground

⚠️ Repo Node.js v1 sudah di-archive di sanhaji182/lintasan

🇬🇧 Migration Guide

Lintasan Go uses a different database schema from Node.js v1 — not backward-compatible.

1. Export settings from Node v1 dashboard
2. Install Lintasan Go (see Installation)
3. Re-create connections in Go v2 dashboard or via API
4. Sync models — auto-discovered on first connection
5. Verify — test via Playground

⚠️ Node.js v1 repo is archived at sanhaji182/lintasan

---

🤝 Contributing

🇮🇩 Kontribusi

Kontribusi sangat diterima! Buka issue untuk bug report atau feature request.

1. Fork repo
2. Buat branch (git checkout -b feature/amazing-feature)
3. Commit changes (git commit -m 'feat: amazing feature')
4. Push ke branch (git push origin feature/amazing-feature)
5. Buka Pull Request

Pastikan tests tetap passing:

go test ./...     # Harus 0 failures

🇬🇧 Contributing

Contributions welcome! Open an issue for bug reports or feature requests.

1. Fork the repo
2. Create a branch (git checkout -b feature/amazing-feature)
3. Commit changes (git commit -m 'feat: amazing feature')
4. Push to branch (git push origin feature/amazing-feature)
5. Open a Pull Request

Make sure tests still pass:

go test ./...     # Must be 0 failures

---

📄 Lisensi / License

MIT

---

🤖 Built with AI

🇮🇩

Project ini dibangun dengan AI sebagai development partner. Arsitektur, keputusan teknis, dan quality control tetap di tangan manusia — AI mengeksekusi.

Lintasan Go adalah rewrite total dari Node.js v1 — mempertahankan semua fitur dengan footprint 35x lebih ringan. Dibuat untuk komunitas AI Indonesia yang percaya bahwa masa depan development adalah orkestrasi, bukan sekadar coding manual.

🇬🇧

This project was built with AI as a development partner. Architecture, technical decisions, and quality control remain in human hands — AI executes.

Lintasan Go is a complete rewrite of Node.js v1 — retaining all features with 35x lighter footprint. Made for the AI community that believes the future of development is orchestration, not just manual coding.

Orchestrator: Sanhaji · Programmer · AI-assisted development advocate

---

🇮🇩 Lintasan Go (v2) — Setiap Koneksi Punya Jalannya
🇬🇧 Lintasan Go (v2) — Every Connection Has Its Path

Dibangun dengan 🤖 AI · Diorkestrasi oleh 👨‍💻 Sanhaji