ennoia 0.1.0

A framework for LLM-powered document pre-indexing and hybrid retrieval.

Ennoia

A framework for LLM-powered document pre-indexing and hybrid retrieval.

Ennoia treats indexing as a first-class problem. Instead of embedding raw text and hoping vector similarity recovers relevance, you declare extraction schemas (Pydantic models for structured metadata, marker classes for semantic summaries), and Ennoia runs an LLM-driven DAG over each document to produce rich, filterable indices. At query time, retrieval runs in two phases: structured filters narrow the candidate set, then vector search ranks within it.

Install

pip install "ennoia[ollama,sentence-transformers,cli]"

Available extras: ollama, openai, anthropic, sentence-transformers, filesystem (Parquet + NumPy stores), cli (ennoia CLI), all (everything above).

Quick start (SDK)

from datetime import date
from typing import Literal

from ennoia import BaseSemantic, BaseStructure, Pipeline, Store
from ennoia.adapters.embedding.sentence_transformers import SentenceTransformerEmbedding
from ennoia.adapters.llm.ollama import OllamaAdapter
from ennoia.store import InMemoryStructuredStore, InMemoryVectorStore


class DocMeta(BaseStructure):
    """Extract basic document metadata."""

    category: Literal["legal", "medical", "financial"]
    doc_date: date


class Summary(BaseSemantic):
    """What is the main topic of this document?"""


pipeline = Pipeline(
    schemas=[DocMeta, Summary],
    store=Store(vector=InMemoryVectorStore(), structured=InMemoryStructuredStore()),
    llm=OllamaAdapter(model="qwen3:0.6b"),
    embedding=SentenceTransformerEmbedding(model="all-MiniLM-L6-v2"),
)

pipeline.index(text="The court held that...", source_id="doc_001")
results = pipeline.search(
    query="court holdings on liability",
    filters={"category": "legal"},
    top_k=5,
)

See docs/quickstart.md for the full walkthrough.

Quick start (CLI)

# Iterate on a schema against a single document
ennoia try ./sample.txt --schema my_schemas.py

# Index a folder into a filesystem-backed store
ennoia index ./docs \
  --schema my_schemas.py \
  --store ./my_index \
  --llm ollama:qwen3:0.6b \
  --embedding sentence-transformers:all-MiniLM-L6-v2

# Hybrid search
ennoia search "employer duty to accommodate disability" \
  --schema my_schemas.py \
  --store ./my_index \
  --filter "jurisdiction=WA" \
  --filter "date_decided__gte=2020-01-01" \
  --top-k 5

See docs/cli.md.

Documentation

• Concepts
• Quickstart
• Schema authoring
• Filter language
• CLI reference
• Adapters
• Stores

License

Apache 2.0. See LICENSE.txt and NOTICE.