Synaplan Multimodal Gateway
공식오픈 소스 서버의 전체 기능을 MCP 예제로 제공합니다.
Synaplan Multimodal Gateway MCP(으)로 무엇을 할 수 있나요?
- Search your RAG knowledge base — 업로드된 문서를 대상으로 질문하고,
POST /mcp엔드포인트를 통해 자체 콘텐츠에 기반한 답변을 얻으세요. - Retrieve AI memories — Synaplan이 캡처한 사용자 프로필과 저장된 컨텍스트를 MCP 서버를 통해 도구 형태로 조회하세요.
- Connect external MCP servers — Channels 아래에 자체 MCP 호환 서비스(CRM, wiki, n8n)를 추가하여 AI 플래너가 실시간 데이터를 가져올 수 있도록 하세요.
- Process multi-step requests — 복잡한 작업을 (추출, 요약, 생성) 단계 그래프로 분해하여 실행 중 실시간 진행 카드를 스트리밍하세요.
문서
Synaplan
AI-powered knowledge management with RAG, chat widgets, and multi-channel integration.
Live instance: web.synaplan.com | Docs: docs.synaplan.com | API: Swagger UI
Prerequisites
- Docker + Docker Compose v2 (Docker Desktop on macOS/Windows, or Docker Engine + the Compose plugin on Linux)
- Git
- 8 GB RAM minimum (16 GB recommended for the local-AI standard install)
- ~9 GB free disk for the standard install (~5 GB for minimal)
- Free TCP ports
5173,8000,8082,8025,3307,6333,11435
Apple Silicon (M1–M4) Macs: Synaplan's container images are published for
linux/amd64, so they run under emulation on Apple Silicon. In Docker Desktop → Settings → General, enable "Use Rosetta for x86/amd64 emulation on Apple Silicon" (macOS 13+) for much faster, more stable containers than the default QEMU. Everything works without it — just slower, and the first build takes longer.
Quick Start
git clone https://github.com/metadist/synaplan.git
cd synaplan
docker compose up -d
Open http://localhost:5173 — the UI is ready in ~2 minutes. With the standard install, local Ollama models (gpt-oss:20b, bge-m3, ~14 GB total) continue downloading in the background — chat that uses local AI will start working once that download finishes (docker compose logs -f backend shows progress). For the fastest first experience, use the Minimal install below.
Install Options
| Mode | Command | Size | Best For |
|---|---|---|---|
| Standard | docker compose up -d | ~9 GB | Full features, local AI |
| Minimal | docker compose -f docker-compose-minimal.yml up -d | ~5 GB | Cloud AI only (Groq/OpenAI) |
For the minimal install, set your API key before starting the stack so the first boot already sees it (avoids a restart). Get a free key at console.groq.com:
echo "GROQ_API_KEY=your_key" >> backend/.env
docker compose -f docker-compose-minimal.yml up -d
Already started without a key? Add it and restart the backend:
echo "GROQ_API_KEY=your_key" >> backend/.env && docker compose restart backend
Access
| Service | URL |
|---|---|
| App | http://localhost:5173 |
| API | http://localhost:8000 |
| API Docs | http://localhost:8000/api/doc |
| phpMyAdmin | http://localhost:8082 |
| MailHog | http://localhost:8025 |
Default Login Credentials:
| Password | Level | |
|---|---|---|
| admin@synaplan.com | admin123 | ADMIN |
| demo@synaplan.com | demo123 | PRO |
| test@example.com | test123 | NEW (unverified) |
Features
- AI Chat — Ollama, OpenAI, Anthropic, Groq, Gemini
- Multi-Task Routing — An AI planner decomposes complex requests into a task graph (extract → summarize → generate → reply) and streams live task cards while the steps execute
- RAG Search — Semantic document search with MariaDB VECTOR or Qdrant
- Chat Widget — Embed on any website (widget guide)
- Live Support — Realtime WebSocket layer (Centrifugo + Redis): human takeover of widget chats, typing indicators, operator notifications (realtime guide)
- WhatsApp — Meta Business API integration
- Email — AI-powered email responses
- Audio — Whisper transcription (input) + optional synaplan-tts (output)
- Documents — PDF, Word, Excel, images with OCR
- AI Memories — User profiling with Qdrant vector search
- Feedback System — Feedback capture and analysis powered by Qdrant
- Plugins — Non-invasive plugin system (plugin guide)
- MCP Server (early access) — Connect AI clients (Claude, Cursor, …) over the Model Context Protocol; your RAG and memories become tools at
POST /mcp(MCP guide) - MCP Client (early access) — Connect your MCP servers (CRM, wiki, n8n, …) under Channels → MCP Servers; the multi-task planner pulls live data from them via
mcp_fetchDAG nodes — read-only, SSRF-guarded, per-topic opt-in. Enabled by seededBCONFIGflags (MCP.CLIENT_ENABLED,MULTITASK.MCP_FETCH_ENABLED—app:seedsets them ON on deploy; an explicit0row is the operator kill switch). See docs/MULTITASK_DATA_NODES.md
Qdrant Vector Database
Qdrant runs as an internal Docker service — no configuration needed. It powers AI memories, RAG document search, and the feedback system.
Starts automatically with docker compose up -d. Synaplan works fully without it (memories and vector search will be disabled).
Realtime & Background Processing
Both compose files also start three internal services (no host ports, no setup needed):
| Service | Role |
|---|---|
redis | Mandatory shared infrastructure: cache, sessions, locks, rate limits, message queues (Redis Streams), Centrifugo engine |
centrifugo | WebSocket gateway for realtime features (live chat takeover, typing indicators, operator notifications) — browsers connect same-origin via /connection/websocket |
worker | Symfony Messenger consumer that executes async jobs (AI processing, document indexing, widget crawling) |
In a multi-node cluster all nodes share one Redis, so WebSocket events published on one node reach browsers connected to any other. Details: docs/REALTIME.md.
Text-to-Speech (Optional)
For voice output, run synaplan-tts alongside Synaplan:
git clone https://github.com/metadist/synaplan-tts.git && cd synaplan-tts && docker compose up -d
Common Commands
# Logs
docker compose logs -f backend
# Restart
docker compose restart backend
# Reset database
docker compose down -v && docker compose up -d
# Run tests
make test
# Code quality
make lint
Documentation
User-facing & API docs live at docs.synaplan.com. Source: metadist/synaplan-docs.
In-repo guides (for developers working on this codebase):
| Guide | Description |
|---|---|
| Installation | Detailed setup instructions |
| Configuration | Environment variables, API keys |
| Development | Commands, testing, architecture |
| Realtime / WebSockets | Centrifugo + Redis realtime layer, multi-node deployment |
| RAG System | Document search and processing |
| Chat Widget | Embed chat on websites |
| Meta Business API setup | |
| Email channel integration |
Related Repositories
| Repo | Purpose |
|---|---|
| synaplan | Main app (this repo) |
| synaplan-docs | Public docs site (docs.synaplan.com) |
| synaplan-tts | Optional Piper TTS service |
| synaplan-sortx | Document-sorting plugin + local tool |
| synaplan-charts | Helm charts for Kubernetes |
| synaplan-platform | Production deployment configs |
Project Structure
synaplan/
├── backend/ # Symfony PHP API
├── frontend/ # Vue.js SPA
├── docs/ # Documentation
├── _docker/ # Docker configs
└── plugins/ # Plugin system
Contributing
See AGENTS.md for development guidelines and code standards.