prismrag-patch 0.1.1

Drop-in hallucination-resistant retrieval for pgvector, ChromaDB, Pinecone, and Weaviate

prismrag-patch

Drop-in hallucination-resistant retrieval for your own vector database.

PrismRAG Patch wraps pgvector, ChromaDB, Pinecone, or Weaviate with PrismRAG's Tier-1 re-mapping technique — deterministic category projection that grounds every chunk in your verified taxonomy before it ever reaches the LLM.

Requirements

Requirement	Detail
Python	3.10+
License key	prlib_ key from prismrag.insightits.com
Embedding model	Gemini gemini-embedding-001 (recommended) or any model producing float vectors
Gemini API key	Required if using Gemini embeddings — get one at aistudio.google.com

Important: prismrag-patch does not generate embeddings for you. You call your embedding model, then pass the resulting vector to the adapter. The library remaps, enriches, and stores/searches the vector in your database.

Quick start — pgvector + Gemini

import os
import requests
import psycopg2
from prismrag_patch import PrismRAGPatch
from prismrag_patch.adapters.pgvector import PgvectorAdapter

# ── 1. Embedding function using Gemini (768-dim, same as production) ─────────
GEMINI_KEY = os.environ["GEMINI_API_KEY"]   # set in your environment

def embed(text: str) -> list[float]:
    """Call Gemini embedding API and return a 768-dim vector."""
    resp = requests.post(
        f"https://generativelanguage.googleapis.com/v1beta/models/"
        f"gemini-embedding-001:batchEmbedContents?key={GEMINI_KEY}",
        json={"requests": [{
            "model": "models/gemini-embedding-001",
            "content": {"parts": [{"text": text}]},
            "outputDimensionality": 768,
            "taskType": "RETRIEVAL_DOCUMENT",
        }]},
        timeout=30,
    )
    resp.raise_for_status()
    return resp.json()["embeddings"][0]["values"]


# ── 2. Define your category mapping ──────────────────────────────────────────
mapping = {
    "categories": [
        {"slug": "risk",   "label": "Risk and Compliance"},
        {"slug": "growth", "label": "Revenue and Growth"},
    ],
    "rules": [
        {"word": "volatility", "category_slug": "risk",   "weight": 1.0},
        {"word": "fraud",      "category_slug": "risk",   "weight": 1.0},
        {"word": "revenue",    "category_slug": "growth", "weight": 1.0},
        {"word": "earnings",   "category_slug": "growth", "weight": 1.0},
    ],
}

# ── 3. Initialize prismrag-patch ──────────────────────────────────────────────
patch = PrismRAGPatch(license_key="prlib_YOUR_KEY_HERE", mapping=mapping)

# ── 4. Connect to your database ───────────────────────────────────────────────
conn    = psycopg2.connect("postgresql://user:pass@localhost:5432/mydb")
adapter = PgvectorAdapter(patch, conn, table="my_chunks")
adapter.ensure_table(dim=768)   # creates table + HNSW index if not exists

# ── 5. Embed your document with Gemini, then insert ──────────────────────────
doc  = "Market volatility spiked due to fraud risk exposure."
vec  = embed(doc)                          # real 768-dim Gemini vector
row_id = adapter.insert(doc, vec, metadata={"source": "risk_report"})
# stored vector is remapped toward "risk" category cluster
# metadata gets prismrag_category + prismrag_label injected automatically

# ── 6. Search ─────────────────────────────────────────────────────────────────
query     = "what is our risk exposure?"
query_vec = embed(query)
results   = adapter.search(query, query_vec, top_k=5)
for r in results:
    print(r["score"], r["metadata"]["prismrag_category"], r["text"][:60])

Batch insert

records = [
    {"text": doc1, "vector": embed(doc1), "metadata": {"source": "q1"}},
    {"text": doc2, "vector": embed(doc2), "metadata": {"source": "q2"}},
]
ids = adapter.batch_insert(records)   # single transaction

Installation

pip install prismrag-patch                   # core only
pip install "prismrag-patch[pgvector]"       # + pgvector (psycopg2)
pip install "prismrag-patch[chroma]"         # + ChromaDB
pip install "prismrag-patch[pinecone]"       # + Pinecone v3
pip install "prismrag-patch[weaviate]"       # + Weaviate v4
pip install "prismrag-patch[all]"            # all adapters

Other adapters

ChromaDB

import chromadb
from prismrag_patch.adapters.chroma import ChromaAdapter

client     = chromadb.Client()
collection = client.get_or_create_collection("my_chunks")
adapter    = ChromaAdapter(patch, collection)
adapter.insert(doc, embed(doc), metadata={"source": "report"})

Pinecone

from pinecone import Pinecone
from prismrag_patch.adapters.pinecone import PineconeAdapter

pc      = Pinecone(api_key=os.environ["PINECONE_API_KEY"])
index   = pc.Index("my-index")
adapter = PineconeAdapter(patch, index, namespace="finance")
adapter.insert(doc, embed(doc), metadata={"source": "report"})

Weaviate

import weaviate
from prismrag_patch.adapters.weaviate import WeaviateAdapter

client     = weaviate.connect_to_local()
collection = client.collections.get("MyChunks")
adapter    = WeaviateAdapter(patch, collection)
adapter.insert(doc, embed(doc))

How it works

Your text
    │
    ▼
Your embedding model (Gemini / OpenAI / Cohere)   ← you call this
    │
    ▼
raw vector  [0.12, -0.45, 0.88, ...]   N dimensions
    │
    ▼
prismrag-patch.insert(text, raw_vector, metadata)
    │
    ├─ category inferred from your rules (local, deterministic, no API call)
    ├─ vector nudged toward winning category cluster (blend_alpha=0.35)
    └─ metadata enriched with prismrag_category + prismrag_label
    │
    ▼
stored in YOUR database — prismrag-patch never holds your data

What prismrag-patch does NOT do

• It does not call an LLM
• It does not generate embeddings
• It does not store your data on PrismRAG servers
• The only network call is a one-time license validation (cached 23 hours, 7-day offline grace period)

License

Commercial license required. Get yours at prismrag.insightits.com/prismrag-lib.html.

© 2026 Insight IT Solutions