Agno Integration

Use blazechunk as an Agno ChunkingStrategy for high-performance semantic chunking.

Installation

Terminal

pip install "blazechunk[agno]"

BlazechunkChunking

BlazechunkChunking is an Agno ChunkingStrategy backed by any blazechunk chunker.

from blazechunk import TokenChunker
from blazechunk.integrations.agno import BlazechunkChunking

# Create a chunking strategy with a custom chunker
strategy = BlazechunkChunking(
    TokenChunker(chunk_size=512, chunk_overlap=64)
)

# Or use the default RecursiveChunker(chunk_size=5000)
default_strategy = BlazechunkChunking()

Methods

from blazechunk import TokenChunker
from blazechunk.integrations.agno import BlazechunkChunking

strategy = BlazechunkChunking(
    TokenChunker(chunk_size=512, chunk_overlap=64)
)

# Chunk a document
document = Document(name="example", page_content="...")
chunked_docs = strategy.chunk(document)  # list[Document]

Using in Agno

Pass the strategy to any Agno knowledge base or reader:

from blazechunk import TokenChunker
from blazechunk.integrations.agno import BlazechunkChunking
from agno.knowledge import TextKnowledgeBase

strategy = BlazechunkChunking(
    TokenChunker(chunk_size=512, chunk_overlap=64)
)

# Use with a knowledge base
knowledge_base = TextKnowledgeBase(
    path="docs",
    vector_db=...,
    chunking_strategy=strategy
)

Note

Each output Document carries the source document's name and metadata, plus a chunk index and chunk_size in its meta_data.