Skip to content

mxLore v2.3.0 — Semantic Search, Auto-Encrypt, Token Efficiency

Choose a tag to compare

View all tags
@MicrotronX MicrotronX released this 08 Apr 13:38
· 308 commits to main since this release
v2.3.0
86e53c8

What's New

Semantic Search

  • Provider-agnostic embedding API (OpenAI, Azure, Ollama, LM Studio — any OpenAI-compatible endpoint)
  • MariaDB VECTOR with VEC_DISTANCE_COSINE for native similarity search
  • Hybrid search: keyword + semantic results merged with configurable weights
  • mx_recall vector boost: lessons ranked by semantic relevance
  • Background timer thread for async embedding (non-blocking boot)
  • Retry with halved input on token limit errors

Auto-Encrypt

  • Plaintext API keys in INI are automatically XOR-encrypted on first boot
  • Supports: Database password, AI API key, Embedding API key

Admin-UI

  • Token Efficiency card: savings %, available vs delivered, MCP calls, sessions
  • Semantic Search card: embedding coverage and stale count

Infrastructure

  • Auto-migrate: detects missing VECTOR column at boot, applies DDL
  • sql/043: VECTOR(1536) column + embedding_stale + triggers
  • Build 78, 45 units, 39 MCP tools

Upgrade from v2.2.1

  1. Replace EXE files
  2. Add to your INI under [AI]:
    EmbeddingEnabled=1
    EmbeddingApiKey=sk-your-openai-key
  3. Restart — server auto-migrates DB schema and starts embedding
Assets 3
Loading