Suggestion: Add dakera-js — AI Agent Memory SDK

[ferhimedamine]

dakera-js

Repository: https://github.com/dakera-ai/dakera-js

Description: TypeScript/JavaScript SDK for Dakera — a self-hosted AI agent memory server. Enables persistent, decay-weighted memory with hybrid search for AI agents.

Category suggestion: Machine Learning / AI or Data Visualization (for knowledge graph)

Key features:

• store() / recall() — persistent agent memories with importance weighting
• hybridSearch() — BM25 + vector ANN combined search
• knowledgeGraph() — entity extraction and graph traversal
• Session lifecycle management
• TypeScript-first with full type definitions
• Works with any AI agent framework (LangChain, Vercel AI SDK, etc.)

Example:

import { DakeraClient } from 'dakera-js';

const client = new DakeraClient({
  apiUrl: 'http://localhost:3300',
  apiKey: 'your-key'
});

// Store a memory
await client.store({
  agentId: 'my-agent',
  content: 'User prefers dark mode and TypeScript examples',
  importance: 0.8,
  tags: ['preference', 'ui']
});

// Hybrid search (BM25 + vector)
const results = await client.hybridSearch({
  namespace: 'my-agent',
  text: 'user preferences',
  topK: 5
});

Server: Self-hosted via Docker (ghcr.io/dakera-ai/dakera:latest)
Deploy: https://github.com/dakera-ai/dakera-deploy
License: MIT