DS DevShelfHub Projects · AI tools
Tutorials / CrewAI / Reference / Classes / ExcelKnowledgeSource
Class knowledge

ExcelKnowledgeSource: Reference Guide

By DevShelfHub

Loads xlsx workbooks (one row per chunk by default).

See the CrewAI API reference index, CrewAI introduction, and core concepts for surrounding context.

What is ExcelKnowledgeSource?

ExcelKnowledgeSource bridges business-owned spreadsheets and CrewAI retrieval without forcing analysts through CSV export rituals. Each row (per sheet) is turned into text for embedding, which preserves human-friendly layouts like merged headers until you normalize them. Multi-sheet workbooks are supported, but remember that sheet order and hidden sheets still consume storage if not filtered upstream.

Unlike CSVKnowledgeSource, Excel carries rich types — dates, currency formats, booleans — that should be materialized as readable strings before embedding. If cells rely on volatile formulas, precompute values offline; otherwise agents retrieve `#REF!` strings that look legitimate in cosine space.

Large color-coded workbooks with thousands of styling-only differences still embed as text. Prefer a clean "data" sheet for retrieval and keep narrative tabs out of the file_path list when possible.

When to Use

Spreadsheet data.

Use Cases

  • • Pricing sheets
  • • FAQ workbooks

Key Features

  • ✓ Per-row embedding
  • ✓ Multi-sheet

When NOT to Use

Heavy formula models — export computed values to CSV.

Notes

Formula freshness

Linked external workbooks and TODAY() cells change meaning after embedding. Snapshot values when generating the xlsx used for RAG.

PII in spreadsheets

Excel is where emails and employee IDs hide in column Q. Run redaction scripts before indexing or scope the source to agents with proper data clearance.

Binary size vs chunk count

A 5 MB xlsx can still explode into tens of thousands of row embeddings. Prune unused sheets and columns before pointing CrewAI at the file.

Comparison to CSVKnowledgeSource

If collaborators already export clean UTF-8 CSVs on a schedule, CSV ingestion is simpler and more diff-friendly in git. Choose Excel when business users insist on xlsx workflows.

Import

python 
from crewai.knowledge.source.excel_knowledge_source import ExcelKnowledgeSource

Key Parameters

Parameter Type Default Purpose
file_path str — Workbook path.

Code Examples

Load a pricing workbook

python 
from crewai.knowledge.source.excel_knowledge_source import ExcelKnowledgeSource

src = ExcelKnowledgeSource(file_path='finance/prices.xlsx')

Agent with Excel + string policy

python 
from crewai import Agent, Task, Crew, Process
from crewai.knowledge.source.excel_knowledge_source import ExcelKnowledgeSource
from crewai.knowledge.source.string_knowledge_source import StringKnowledgeSource

sheet = ExcelKnowledgeSource(file_path='ops/runbook.xlsx')
policy = StringKnowledgeSource(content='Always cite sheet name when answering.')
agent = Agent(role='Ops', goal='Answer runbook questions', backstory='Use both sources.', knowledge_sources=[sheet, policy])

Crew(agents=[agent], tasks=[Task(description='How do we fail over region {region}?', expected_output='Steps', agent=agent)], process=Process.sequential).kickoff(inputs={'region': 'eu-west'})

Crew-wide Excel corpus

python 
from crewai import Crew, Agent, Task, Process
from crewai.knowledge.source.excel_knowledge_source import ExcelKnowledgeSource

src = ExcelKnowledgeSource(file_path='catalog/products.xlsx')
writer = Agent(role='Writer', goal='Draft copy', backstory='Ground claims in catalog rows.')

Crew(agents=[writer], tasks=[Task(description='Describe SKU {sku}', expected_output='Marketing blurb', agent=writer)], knowledge_sources=[src], process=Process.sequential).kickoff(inputs={'sku': 'SKU-900'})

Common Mistakes

❌ Indexing a template workbook with empty example rows

✅ Filter to populated rows or maintain a dedicated data sheet without placeholders.

ExcelKnowledgeSource FAQ

What is ExcelKnowledgeSource in CrewAI?

Loads xlsx workbooks (one row per chunk by default). ExcelKnowledgeSource bridges business-owned spreadsheets and CrewAI retrieval without forcing analysts through CSV export rituals. Each row (per sheet) is turned into text for embedding, which preserves human-friendly layouts like merged headers until you normalize them. Multi-sheet workbooks are supported, but remember that sheet order and hidden sheets still consume storage if not filtered upstream. Unlike CSVKnowledgeSource, Excel carries rich types — dates, currency form…

Which package defines the CrewAI class ExcelKnowledgeSource?

DevShelfHub maps ExcelKnowledgeSource to Python module crewai.knowledge.source (package path crewai.knowledge.source in this reference). Pin your installed crewai version and match imports to the snippet on this page.

When should I use ExcelKnowledgeSource?

Spreadsheet data.

When should I avoid using ExcelKnowledgeSource?

Heavy formula models — export computed values to CSV.

How do I import ExcelKnowledgeSource in Python?

from crewai.knowledge.source.excel_knowledge_source import ExcelKnowledgeSource

Where can I explore more CrewAI API reference pages?

Open the CrewAI API reference index on DevShelfHub to search 58 classes, 30 methods, and 16 decorators, each with runnable examples, parameters, common mistakes, and cross-links.