Repository navigation
v0.9.0
What's New in v0.9.0
Progressive Disclosure CLI
loom setup— interactive wizard, auto-detects Ollama, writes ~/.loom/config.yamlloom new worker/loom new pipeline— interactive config scaffoldingloom validate— check configs without starting infrastructureloom rag ingest/search/stats/serve— zero-config RAG pipeline
Ready-Made Workers
Six shipped workers, usable immediately or as pipeline building blocks:
- summarizer, classifier, extractor (existing)
- translator — multi-language with auto-detection
- qa — question answering with source citations (completes the RAG story)
- reviewer — quality review against configurable criteria
New Contrib Modules
- contrib/docproc — document extraction (MarkItDown fast path, Docling OCR, SmartExtractor fallback)
- contrib/lancedb — ANN vector store for large-scale retrieval
RAG Pipeline Enhancements
- Ingestor/VectorStore ABCs for pluggable backends
- Live Telegram channel monitoring via Telethon
- Workshop RAG dashboard (store stats, channel registry, semantic search)
PyPI Publishing
- Install:
pip install loom-ai(orpip install loom-ai[rag],loom-ai[workshop]) - Import name unchanged:
import loom
Stats
- 1643 tests, 90% coverage
- 19 CLI commands
- 107 Python modules