Skip to content

📚 OpenWiki.network

Part of Ecosystem: OpenOS

The OpenWiki.network is a distributed collective knowledge graph system for AI, cognitive architectures, agents and applications.

OpenWiki.Network provides hybrid knowledge store that unifies structured multimodal storage (Documents, KV store, Graphs & more) with semantic embeddings (Vector store), along with ingestion pipelines and Kubernetes deployers - giving you a complete foundation for open, distributed network wide reasoning, planning and search.

OpenWiki.Network enables collective knowledge contribution, inference & enrichment.

Kubernetes Deployment Layer

Inside wiki_stack_deployer/core/ are Python-based deployers for Kubernetes-native installation:

  • arango_deployer.py → Deploys ArangoDB clusters with replication, sharding, TLS, and authentication.
  • embeddings_deployer.py → Deploys Weaviate as a StatefulSet or provisions Pinecone indexes.
  • api.py → Flask REST API layer to deploy ArangoDB, Embeddings DB, or both in one call.
  • k8s/deploy.yaml + k8s/rbac.yaml → Core manifests for stack installation and RBAC setup.

This forms the in-cluster data backbone for agents.

SDK Layer

Inside wiki_sdk/agents_wiki/ are Python SDKs to interact directly with databases and embeddings:

  • Arango SDK (db.py)
  • Full CRUD (insert, update, delete, query)
  • Database & collection lifecycle management

  • Ingest JSON/JSONL data from local files or S3

  • Embeddings SDK (embeddings.py)

  • Unified interface for Weaviate & Pinecone
  • Vector CRUD (insert, update, delete, fetch)
  • 🔍 Search (nearest neighbor + filters)

  • Index/collection lifecycle management

  • Embeddings Store (embeddings_store.py)

  • Arango → Embeddings ingestion bridge
  • Abstract EmbeddingModel class for custom embedding models (OpenAI, Hugging Face, etc.)
  • Batched streaming with optional L2 normalization and metadata mapping

These SDKs allow agents to persist, embed, and search knowledge with minimal integration work.

System Services Layer

The stack also includes supporting services and runtime components:

  • Dockerfile (wiki_stack_deployer/) → Build deployer as a containerized service.
  • requirements.txt files → Manage dependencies for SDK and deployer separately.
  • RBAC & Deploy manifests (k8s/) → Kubernetes-native configuration for secure deployment.

This ensures the system is ready for production, with pluggable deployers and SDKs.

🎯 Supported Components

Subsystem Brief
K8s Deployers Python-based deployers for ArangoDB and Embeddings DB (Weaviate/Pinecone).
Arango SDK CRUD, lifecycle management, ingestion (JSON, JSONL, S3).
Embeddings SDK Unified interface for Weaviate & Pinecone, with search & CRUD.
Embeddings Store Ingestor: Arango → Embeddings DB, pluggable embedding models.
Flask API Layer REST APIs to deploy Arango, Embeddings DB, or both in one call.
K8s Manifests Deployment YAMLs + RBAC setup for Kubernetes clusters.

🚀 Features

  • 📚 ArangoDB SDK for structured graph + document CRUD
  • 🔍 Embeddings SDK with nearest-neighbor search & metadata filters
  • 🧠 Arango → Embeddings ingestion with pluggable models (OpenAI, Hugging Face, custom)
  • ☁️ Kubernetes-native deployers for Arango & Embeddings DBs
  • 🔑 Secure deployments with RBAC, TLS, and authentication options
  • 🛠 Extensible architecture: swap embedding backends, models, or storage layers

🤝 Contributing

This project is community-driven. Contributions welcome!

Ways to Contribute

  • Add new embedding backends (Milvus, FAISS, Qdrant)
  • Extend ingestion with incremental sync & tombstone deletes
  • Build CLI tools for quick deployment & ingestion
  • Add monitoring & dashboards for deployed stacks
  • Write tutorials & real-world examples

📢 Communications