Pluggedin App

What is Pluggedin App?

Pluggedin App is the world's first AI Content Management System (AI-CMS) that transforms ephemeral AI conversations into permanent, versioned, and searchable organizational memory. It integrates with 1,500+ MCP servers and supports multiple AI models (Claude, GPT-4, Gemini) to store, attribute, and retrieve AI-generated content.

How to use Pluggedin App?

Self‑host via Docker using docker compose up --build -d (requires Docker with BuildKit and ~8 GB RAM for the initial build), or use the cloud version at plugged.in with zero installation. After setup, configure MCP servers, upload documents, and interact via the web UI or SDKs (JavaScript/TypeScript, Python, Go).

Key features of Pluggedin App

• AI memory persistence with full versioning and model attribution
• Multi‑model collaboration (Claude, GPT‑4, Gemini) in one document
• Universal MCP integration with 1,500+ servers
• Embedded RAG vector engine (zvec) – no external services
• Git‑style document version control
• Enterprise‑grade security (AES‑256‑GCM, OAuth 2.1, sandboxed execution)

Use cases of Pluggedin App

• Store and search AI‑generated analyses, code reviews, and strategy documents
• Track contributions from multiple AI models per document (e.g., Claude v1, GPT‑4 v2)
• Unify 1,500+ MCP tools through a single proxy interface
• Self‑host a private AI knowledge base with built‑in semantic search
• Use the clipboard/memory system for temporary data with TTL and MCP integration

FAQ from Pluggedin App

What is the difference between the self‑hosted and cloud versions?

Self‑hosted builds the app image from source (due to the embedded zvec engine) and requires Docker with BuildKit and ~8 GB RAM. The cloud version at plugged.in is instantly accessible with no installation – a prebuilt production image is used.

What are the system requirements for self‑hosting?

Docker with BuildKit (Docker Desktop or docker buildx), at least ~8 GB of memory allocated to the Docker engine for the next build, and the bundled PostgreSQL 18 with pgvector and Redis 7 images.

How does the RAG engine work?

The RAG engine is an embedded zvec vector store (RocksDB + HNSW) that runs fully in‑process – document processing, chunking, and semantic search all happen inside the application with no external Milvus or Qdrant dependencies.

Can I access Pluggedin App programmatically?

Yes. Official SDKs are available for JavaScript/TypeScript (@pluggedin/sdk), Python (pluggedin-sdk), and Go (pluggedin-go) to create documents, query the RAG knowledge base, and interact with MCP servers via code.

Is there a clipboard or temporary memory system?

Yes. The system supports named key‑value entries with TTL expiration, stack operations (push/pop), and visibility controls (private, workspace, public). It is accessible via 6 built‑in MCP tools from any AI client.