SDK Comparison

SDK Feature Comparison

Feature availability across DeltaMemory SDKs and integrations.

Core SDKs

FeatureTypeScriptPythonStatus
Core Operations
ingest()Stable
recall()Stable
store()Stable
get()Stable
delete()Stable
Cognitive Features
User ProfilesStable
Event TimelineStable
Pre-formatted ContextStable
Fact ExtractionStable
Concept ExtractionStable
Memory Management
decay()Stable
consolidate()Stable
reflect()Stable
purge()Stable
Knowledge Graph
graph()Stable
Multi-hop ReasoningStable
Collections
Multiple CollectionsStable
Collection StatsStable
Advanced
Custom WeightsStable
Memory Type FilteringStable
Metadata SupportStable
Batch Operations⚠️⚠️Planned
StreamingPlanned
Developer Experience
TypeScript TypesN/AStable
Python Type HintsN/AStable
Error ClassesStable
Async/AwaitStable
Context ManagerN/AStable

Integration SDKs

FeatureAI SDKMCPStatus
Core Operations
Recall MemoryStable
Store MemoryStable
Get ProfileStable
Get EventsStable
Framework Support
Vercel AI SDKN/AStable
Claude DesktopN/AStable
ClineN/AStable
Other MCP ClientsN/AStable
Features
Tool DescriptionsStable
Zod SchemasN/AStable
Streaming SupportStable
Multi-user IsolationStable
Custom MetadataStable

Language Support

LanguagePackageVersionStatus
TypeScript/JavaScriptdeltamemory1.0.0+✅ Stable
Pythondeltamemory1.0.0+✅ Stable
Go--🔮 Planned
Rust--🔮 Planned
Java--🔮 Planned

Framework Integrations

FrameworkPackageStatus
Vercel AI SDK@deltamemory/ai-sdk✅ Stable
LangChain (Python)deltamemory-langchain⚠️ Beta
LangChain (JS)-🔮 Planned
LlamaIndex-🔮 Planned
Haystack-🔮 Planned
Semantic Kernel-🔮 Planned

Platform Support

PlatformTypeScriptPythonNotes
Node.js 18+N/ARecommended
Node.js 16+N/ASupported
BunN/ASupported
Deno⚠️N/AUntested
Python 3.8+N/ARecommended
Python 3.7N/AUnsupported
Browser⚠️N/AAPI key exposure risk

Deployment Platforms

PlatformSupportNotes
VercelServerless functions
AWS LambdaConnection reuse recommended
Google Cloud FunctionsConnection reuse recommended
Azure FunctionsConnection reuse recommended
DockerFull support
KubernetesFull support
Cloudflare Workers⚠️Limited runtime

Legend

  • Stable - Production ready, fully supported
  • ⚠️ Beta - Available but may have breaking changes
  • 🔮 Planned - On roadmap, not yet available
  • Not Available - Not currently supported
  • N/A - Not applicable for this SDK

Version Requirements

TypeScript SDK

{
  "node": ">=16.0.0",
  "typescript": ">=4.5.0"
}

Python SDK

[tool.poetry.dependencies]
python = "^3.8"
httpx = "^0.24.0"

AI SDK Integration

{
  "node": ">=18.0.0",
  "ai": ">=3.0.0",
  "deltamemory": ">=1.0.0"
}

Migration Path

From CognitiveDB to DeltaMemory

Both TypeScript and Python SDKs support legacy CognitiveDB alias:

// Still works
import { CognitiveDB } from 'deltamemory';
const db = new CognitiveDB({ ... });
# Still works
from deltamemory import CognitiveDB
db = CognitiveDB(...)

Recommendation: Migrate to DeltaMemory class name for future compatibility.

Feature Requests

Missing a feature? Let us know:

Roadmap

Q1 2024

  • ✅ User profiles and events
  • ✅ Pre-formatted context
  • ✅ MCP server integration

Q2 2024

  • ⚠️ LangChain Python integration
  • 🔮 Batch operations API
  • 🔮 Streaming responses

Q3 2024

  • 🔮 LangChain JS integration
  • 🔮 LlamaIndex integration
  • 🔮 Go SDK

Q4 2024

  • 🔮 Rust SDK
  • 🔮 Java SDK
  • 🔮 Advanced graph queries

Roadmap subject to change based on user feedback and priorities.

SDK Comparison Details

TypeScript SDK

Strengths:

  • Full TypeScript type safety
  • Native async/await
  • Works with all Node.js frameworks
  • Bun support

Best for:

  • Next.js applications
  • Express/Fastify APIs
  • Serverless functions
  • Real-time applications

Python SDK

Strengths:

  • Pythonic async/await
  • Context manager support
  • Type hints with mypy
  • httpx-based (modern, fast)

Best for:

  • FastAPI applications
  • Django applications
  • Data science workflows
  • ML/AI pipelines

AI SDK Integration

Strengths:

  • Native Vercel AI SDK tools
  • Streaming support
  • Zod schema validation
  • Agent-friendly

Best for:

  • Vercel AI SDK projects
  • Chatbot applications
  • Agent-based systems
  • Streaming UIs

MCP Server

Strengths:

  • Works with any MCP client
  • Claude Desktop integration
  • Cline integration
  • No code changes needed

Best for:

  • Claude Desktop users
  • IDE integrations
  • Development tools
  • Quick prototyping

Performance Characteristics

OperationTypeScriptPythonNotes
ingest()~500-1500ms~500-1500msLLM processing time
store()~50-200ms~50-200msEmbedding only
recall()~100-500ms~100-500msDepends on collection size
get()~20-50ms~20-50msDirect lookup
graph()~100-300ms~100-300msDepends on graph size

Performance varies based on content size, collection size, and network latency.

Support Matrix

SDKDocumentationExamplesTestsCI/CD
TypeScript
Python
AI SDK⚠️
MCP⚠️

Questions?