orbmem 1.1.1

OrbMem – A lightweight Cognitive Memory & Vector framework (Memory, Vectors, Graph, Safety).

OCDB — Cognitive Database Framework

OCDB (Orbynt Cognitive Database) is a lightweight, modular framework that combines 🧠 Memory, 🔍 Vector Search, 🕸 Graph Reasoning, and 🛡 Safety Scanning into a single Python package.

Designed for:

• AI agents
• Chatbots
• Automation systems
• Reasoning pipelines
• Personal assistants
• Any system that needs context, memory, search, and safety

This is the local developer version for PyPI. A fully hosted OCDB Cloud Platform will launch separately.

---

✨ Features

🧠 1. Memory Engine

Store and retrieve structured memory using a simple API.

from orbmem.core.ocdb import OCDB

ocdb = OCDB()
ocdb.memory_set("name", {"value": "Abhishek"})
print(ocdb.memory_get("name"))

---

🔍 2. Vector Search (FAISS local engine)

Semantic search powered by FAISS.

ocdb.vector_engine.add_text("hello world", {"id": "1"})
results = ocdb.vector_search("hello", k=3)

---

🕸 3. Graph Engine (NetworkX)

Build reasoning trees or node-based logic.

ocdb.graph_add("root", "Start")
ocdb.graph_add("child", "Next", parent="root")

---

🛡 4. Safety Engine

Simple text-safety scanning with MongoDB + time-series tracking.

events = ocdb.safety_scan("This is a sample text")

---

📦 Installation

pip install orbmem

---

📁 Project Structure

ocdb/
 ├── core/        # Config, auth, high-level OCDB interface
 ├── engines/     # Memory, vector, graph, safety engines
 ├── models/      # Memory & safety models
 ├── utils/       # Helpers, exceptions, logging
 └── db/          # Database connectors (Mongo, Neo4j, Postgres, Redis)

---

🚀 Quickstart Example

from orbmem.core.ocdb import OCDB

ocdb = OCDB()

# Memory
ocdb.memory_set("demo", {"text": "Hello OCDB"})
print(ocdb.memory_get("demo"))

# Vector
print(ocdb.vector_search("Abhishek"))

# Graph
ocdb.graph_add("n1", "Start")
ocdb.graph_add("n2", "Next", parent="n1")

# Safety
print(ocdb.safety_scan("Harmless text"))

Install dependencies:

pip install -r requirements.txt

From source:

git clone https://github.com/abhis-byte/orbynt-database-OrbMem.git
cd orbmem
pip install .
pip install -r requirements.txt

---

🚀 Basic Usage

from orbmem.core.ocdb import OCDB 
db = OCDB()

This initializes:

• SQLite memory backend
• NetworkX graph backend
• SQLite safety backend (unless MONGO_URL is set)
• FAISS fallback vector backend
• Time-series safety tracking

---

🧠 Memory Engine (SQLite)

Store memory:

db.memory_set("user", {"name": "Abhishek"})

Retrieve memory:

print(db.memory_get("user"))

List keys:

print(db.memory_keys())

The memory engine is SQLite-based and works everywhere without configuration.

---

🔗 Graph Engine (NetworkX fallback)

Add graph nodes:

db.graph_add("root", "This is the root node") 
db.graph_add("child", "This is a child node", parent="root")

Get relationship path:

print(db.graph_path("root", "child"))

Dump full graph:

print(db.graph_dump())

The graph engine automatically uses NetworkX unless Neo4j configuration is provided.

---

🔍 Vector Engine (FAISS Fallback)

Version 1 uses a lightweight FAISS fallback for simple vector search examples.

Future versions (v2) will include:

• Real embeddings (SentenceTransformers)
• Qdrant/FAISS hybrid vector search
• Cloud embedding cache
• High-performance ANN indexing

Current example:

results = db.vector_search("hello world", k=3)
print(results)

---

🛡 Safety Engine (SQLite / Mongo)

The safety engine detects unsafe text patterns and records violations.

Scan text:

db.safety_scan("This text contains harmful intent.")
print(events)

Safety results contain:

• text
• tag
• severity
• correction
• details
• timestamp

Example output:

[
  {
    "text": "This text contains harmful intent.",
    "tag": "violence",
    "severity": 0.6,
    "correction": "Avoid harmful or violent expressions.",
    "details": {"pattern": ""},
    "timestamp": "2025-01-01T12:00:00Z"
  }
]

Backend selection:

• If MONGO_URL is set → MongoSafetyBackend
• Else → SQLiteSafetyBackend (portable)

---

📘 Example Code (Quick Demo)

from orbmem.core.ocdb import OCDB

db = OCDB()

# Memory
db.memory_set("user", {"name": "Abhishek"})
print(db.memory_get("user"))

# Vector search
print(db.vector_search("hello", k=3))

# Graph
db.graph_add("root", "Root node")
db.graph_add("child", "Child node", parent="root")
print(db.graph_path("root", "child"))
print(db.graph_dump())

# Safety engine
db.safety_scan("This may be violent content")

---

📌 Notes

• Designed for local usage (v1)
• Cloud support coming in v2
• SQLite-only mode works in all environments:
  • Kaggle
  • Jupyter
  • Colab
  • VSCode
• Windows/macOS/Linux
• No environment variables required for v1

---

💙 ORBMEM Cloud (Coming in v2)

• Real SentenceTransformer embeddings
• Qdrant / FAISS hybrid vector engine
• Neo4j cloud graph reasoning
• MongoDB safety event storage
• Postgres memory engine
• API keys
• Developer dashboard
• Remote sync

Stay tuned.

---

📝 License

This project is released under the MIT License.
See the LICENSE file for full terms.

---

👨‍💻 Author

Tetala Lakshmi Abhishek Reddy
Creator of OCDB
📧Email: [e-mail]

---

⭐ Support OCDB

If you like this project, please star the repository and share it!
The OCDB Cloud Dashboard + Hosted API is coming soon… 🚀