Movatterモバイル変換

CoderCouple/context0-python-backendPublic

NotificationsYou must be signed in to change notification settings
Fork0
Star0

You must be signed in to change notification settings

Folders and files

Name		Name	Last commit message	Last commit date
Latest commit History 14 Commits
.idea		.idea
app		app
docker-init-scripts		docker-init-scripts
docs		docs
wiki		wiki
.env.example		.env.example
.gitignore		.gitignore
.pre-commit-config.yaml		.pre-commit-config.yaml
Dockerfile		Dockerfile
Makefile		Makefile
README.md		README.md
context0_app.py		context0_app.py
create_indexes.py		create_indexes.py
docker-compose.yml		docker-compose.yml
generate_test_token.py		generate_test_token.py
ngrok.yml		ngrok.yml
poetry.lock		poetry.lock
pyproject.toml		pyproject.toml
start-ngrok.sh		start-ngrok.sh

Repository files navigation

make a# Context0 Memory System

A sophisticated AI memory management system built with FastAPI that provides intelligent memory storage, retrieval, and time-travel debugging capabilities.

Features

🧠 Memory Types

Semantic: Knowledge and facts
Episodic: Personal experiences and events
Procedural: Skills and how-to knowledge
Emotional: Feelings and emotional context
Working: Temporary active information
Declarative: Explicit knowledge
Meta: Knowledge about knowledge

🚀 Core Capabilities

LLM-Powered Classification: Automatic memory type inference using OpenAI/Anthropic
Multi-Store Architecture: Vector, Graph, Document, TimeSeries, and Audit stores
Time-Travel Debugging: Navigate memory state at any point in time
Async Processing: Full async/await support for high performance
Rich Metadata: Comprehensive tagging, permissions, and relationships

🏗️ Architecture

MemoryRecord → MemoryTypeInferencer → MemoryRouter → Handler.process() → MemoryEntry → Stores

Quick Start

Prerequisites

Python 3.11+
Poetry
Required external services (Pinecone, MongoDB, Neo4j, TimescaleDB)

Installation

# Clone the repositorygit clone<repository-url>cd context0-python-backend# Install dependenciespoetry install# Install additional memory system dependenciespoetry add pinecone-client motor pymongo neo4j asyncpg langchain openai python-dateutil

Configuration

Set environment variables for external services:

# OpenAIexport OPENAI_API_KEY="your-openai-key"# Pineconeexport PINECONE_API_KEY="your-pinecone-key"export PINECONE_ENVIRONMENT="your-environment"# MongoDBexport MONGODB_URL="mongodb://localhost:27017"# Neo4jexport NEO4J_URI="bolt://localhost:7687"export NEO4J_USER="neo4j"export NEO4J_PASSWORD="password"# TimescaleDBexport TIMESCALE_URL="postgresql://user:pass@localhost:5432/db"

Running the Application

# Development server with auto-reloadpoetry run python context0_app.py app.main:app --reload# Or using uvicorn directlypoetry run uvicorn app.main:app --reload --host 0.0.0.0 --port 8000

The API will be available athttp://localhost:8000

Memory System Architecture

Models

MemoryRecord

Runtime context for incoming observations:

classMemoryRecord(BaseModel):user_id:strsession_id:Optional[str]memory_type:Optional[MemoryType]raw_text:strtimestamp:datetimetags:List[str]metadata:Dict[str,Any]

MemoryEntry

Canonical persisted memory object with rich metadata:

classMemoryEntry(BaseModel):id:strcid:str# Content ID for deduplicationscope:strinput:strsummary:Optional[str]memory_type:MemoryTypepermissions:MemoryPermissionsembedding:Optional[List[float]]graph_links:List[GraphLink]

Storage Backends

Pinecone: Vector embeddings for semantic search
Neo4j: Knowledge graphs and relationships
MongoDB: Document storage with full-text search
TimescaleDB: Temporal analysis and time-series data
MongoDB Audit: Time-travel debugging and state snapshots

Handlers

Specialized processors for each memory type:

SemanticHandler: Extracts embeddings and knowledge
EpisodicHandler: Temporal context and experiences
ProceduralHandler: Step-by-step processes
EmotionalHandler: Sentiment and emotional context
WorkingHandler: Active information management
DeclarativeHandler: Explicit facts
MetaHandler: Self-referential knowledge

API Endpoints

Memory Operations

POST /api/v1/memory/: Create new memory
GET /api/v1/memory/{memory_id}: Retrieve memory
GET /api/v1/memory/search: Search memories
GET /api/v1/memory/time-travel: Time-travel queries

Analytics

GET /api/v1/analytics/: System analytics
GET /api/v1/analytics/memory-stats: Memory statistics

Health & Monitoring

GET /api/v1/ping: Health check
GET /api/v1/webhook: Webhook management

Time-Travel Debugging

Query memory state at any point in time:

# Get all memories that existed at a specific timememories=awaitaudit_store.get_memory_state_at_time(user_id="user123",target_time=datetime(2024,1,15,10,30))# Trace memory evolutionevolution=awaitaudit_store.get_memory_evolution(memory_id="mem_456",start_time=start_time,end_time=end_time)

Development

Code Style

Black: Code formatting
isort: Import sorting
mypy: Type checking
pre-commit: Git hooks

Testing

poetry run pytest

Pre-commit Hooks

poetry run pre-commit installpoetry run pre-commit run --all-files

Project Structure

app/├── memory/                   # Memory system core│   ├── models/               # Data models│   ├── inferencer/           # LLM-based type inference│   ├── router/               # Memory routing logic│   ├── handlers/             # Type-specific processors│   ├── stores/               # Storage backends│   ├── engine/               # Main orchestrator│   └── config/               # Configuration management├── api/                      # FastAPI routes├── common/                   # Shared utilities├── service/                  # Business logic└── model/                    # Database models

Contributing

Fork the repository
Create a feature branch
Make your changes
Run tests and linting
Submit a pull request

License

[License details here]

Support

For questions and support, pleasecreate an issue in the repository.

About

No description, website, or topics provided.

Releases

1tags

Packages

No packages published

Languages

Python99.8%
Other0.2%