Skip to content

Releases: two-tech-dev/memorable

v0.2.0

17 Apr 06:21
@0xliam627 0xliam627
v0.2.0
7de940a

Choose a tag to compare

View all tags
v0.2.0 Latest
Latest

Memorable v0.2.0

Long-term memory engine for AI agents. Give your coding assistant a brain that persists across sessions.

Highlights

  • 12 MCP tools — Full memory CRUD, semantic search, heartbeat consolidation, and knowledge graph via the Model Context Protocol
  • 4 embedding providers — OpenAI, Google Gemini, Ollama (local), and any OpenAI-compatible API (OpenRouter, Together, Voyage, Mistral, Cohere, Azure)
  • PostgreSQL + pgvector — Production-grade vector storage with cosine similarity search
  • Auto-install scripts — One command to configure Cursor, Claude Code, GitHub Copilot, and Windsurf

Features

Memory Management

  • Store, search, update, delete memories with semantic vector search
  • 5 memory types: fact, conversation, decision, code_pattern, correction
  • SHA-256 content deduplication within scopes
  • Multi-dimensional scoping: user × agent × app × run
  • Pagination and filtering for memory listing

Knowledge Graph

  • Automatic entity-relation extraction from text
  • Entity search by name with fuzzy matching
  • Graph neighbor traversal with configurable depth
  • Provenance tracking via memory IDs

Heartbeat / Self-Reflection

  • Periodic memory consolidation into insights
  • Cluster detection for related memories
  • Contradiction detection between corrections and older facts
  • Pattern recognition across stored memories

Hybrid Retrieval

  • Vector similarity + recency + frequency scoring
  • Configurable scoring weights
  • L1 LRU cache for hot queries

Architecture

  • L1 Cache (thread-safe LRU) → L2 Vector DB (pgvector) → L3 Soul Profile (trait accumulation)
  • Clean internal/ package structure with well-defined interfaces
  • MCP SDK v1.5.0 with typed tool handlers and auto-generated JSON schemas

Supported Agents

  • Cursor
  • Claude Desktop / Claude Code
  • VS Code (GitHub Copilot)
  • Windsurf
  • Antigravity

Embedding Providers

Provider Models Local
OpenAI text-embedding-3-small, text-embedding-3-large No
Google Gemini text-embedding-004 No
Ollama nomic-embed-text, mxbai-embed-large, all-minilm Yes
Custom Any OpenAI-compatible /v1/embeddings API Varies

Getting Started

git clone https://github.com/two-tech-dev/memorable.git
cd memorable && make build
install.sh  # auto-configure your AI agent
Assets 2
Loading