Knowledge Sources

A knowledge source is a concrete AgentKnowledge subclass that knows how to ingest one kind of input. Each lives in its own module under buddy.knowledge and ships with a default reader, so you only provide the source (a path, urls, queries, topics, …) and a vector_db.

Verified class names

These names were read from buddy/knowledge/. Note UrlKnowledge (not UrlKnowledgeBase) and the lowercase irag. The base class AgentKnowledge is importable as from buddy.knowledge import AgentKnowledge.

File-based sources

Source	Import	Class	Main field
PDF	`from buddy.knowledge.pdf import PDFKnowledgeBase`	`PDFKnowledgeBase`	`path`
Text	`from buddy.knowledge.text import TextKnowledgeBase`	`TextKnowledgeBase`	`path`
CSV	`from buddy.knowledge.csv import CSVKnowledgeBase`	`CSVKnowledgeBase`	`path`
JSON	`from buddy.knowledge.json import JSONKnowledgeBase`	`JSONKnowledgeBase`	`path`
DOCX	`from buddy.knowledge.docx import DocxKnowledgeBase`	`DocxKnowledgeBase`	`path`
Markdown	`from buddy.knowledge.markdown import MarkdownKnowledgeBase`	`MarkdownKnowledgeBase`	`path`
PDF bytes	`from buddy.knowledge.pdf_bytes import PDFBytesKnowledgeBase`	`PDFBytesKnowledgeBase`	`pdfs`

Web & URL sources

Source	Import	Class	Main field
URL	`from buddy.knowledge.url import UrlKnowledge`	`UrlKnowledge`	`urls`
Website (crawl)	`from buddy.knowledge.website import WebsiteKnowledgeBase`	`WebsiteKnowledgeBase`	`urls`
PDF URL	`from buddy.knowledge.pdf_url import PDFUrlKnowledgeBase`	`PDFUrlKnowledgeBase`	`urls`
CSV URL	`from buddy.knowledge.csv_url import CSVUrlKnowledgeBase`	`CSVUrlKnowledgeBase`	`urls`
Firecrawl	`from buddy.knowledge.firecrawl import FireCrawlKnowledgeBase`	`FireCrawlKnowledgeBase`	`urls`

Research & media sources

Source	Import	Class	Main field
arXiv	`from buddy.knowledge.arxiv import ArxivKnowledgeBase`	`ArxivKnowledgeBase`	`queries`
Wikipedia	`from buddy.knowledge.wikipedia import WikipediaKnowledgeBase`	`WikipediaKnowledgeBase`	`topics`
YouTube	`from buddy.knowledge.youtube import YouTubeKnowledgeBase`	`YouTubeKnowledgeBase`	`urls`

Composite & integration sources

Source	Import	Class	Notes
Combined	`from buddy.knowledge.combined import CombinedKnowledgeBase`	`CombinedKnowledgeBase`	`sources: List[AgentKnowledge]`
Documents	`from buddy.knowledge.document import DocumentKnowledgeBase`	`DocumentKnowledgeBase`	In-memory `Document`s
LangChain	`from buddy.knowledge.langchain import LangChainKnowledgeBase`	`LangChainKnowledgeBase`	Wraps a LangChain retriever/vectorstore
LlamaIndex	`from buddy.knowledge.llamaindex import LlamaIndexKnowledgeBase`	`LlamaIndexKnowledgeBase`	Wraps a LlamaIndex `retriever`
LightRAG	`from buddy.knowledge.light_rag import LightRagKnowledgeBase`	`LightRagKnowledgeBase`	Talks to a LightRAG server
iRAG	`from buddy.knowledge import irag`	`irag`	Local SQLite RAG — see RAG

Examples

PDF fileWebsiteWikipediaCombined

from buddy.knowledge.pdf import PDFKnowledgeBase
from buddy.vectordb.chroma import ChromaDb

kb = PDFKnowledgeBase(path="docs/handbook.pdf",
                      vector_db=ChromaDb(collection="handbook"))
kb.load()

from buddy.knowledge.website import WebsiteKnowledgeBase
from buddy.vectordb.chroma import ChromaDb

kb = WebsiteKnowledgeBase(
    urls=["https://example.com/docs"],
    max_depth=3, max_links=10,
    vector_db=ChromaDb(collection="site"),
)
kb.load()

from buddy.knowledge.wikipedia import WikipediaKnowledgeBase
from buddy.vectordb.chroma import ChromaDb

kb = WikipediaKnowledgeBase(topics=["Quantum computing"],
                            vector_db=ChromaDb(collection="wiki"))
kb.load()

from buddy.knowledge.combined import CombinedKnowledgeBase

kb = CombinedKnowledgeBase(sources=[pdf_kb, website_kb], vector_db=shared_db)
kb.load()

Optional dependencies

Some sources pull in extra packages — e.g. WikipediaKnowledgeBase requires wikipedia, LlamaIndexKnowledgeBase requires llama-index-core, and FireCrawlKnowledgeBase requires the Firecrawl client. A missing dependency raises an ImportError on import.