Framework Integrations

LangChain

Five StructuredTools wrapping the most useful free TensorFeed endpoints, plus a Document loader for AI news. Pass tensorfeed_tools() directly into a LangGraph agent.

Install

pip install 'tensorfeed[langchain]'

Use as agent tools (LangGraph)

from langchain_openai import ChatOpenAI
from langgraph.prebuilt import create_react_agent
from tensorfeed.langchain import tensorfeed_tools

agent = create_react_agent(
    ChatOpenAI(model="gpt-5.5"),
    tensorfeed_tools(),
)

result = agent.invoke({
    "messages": [("user", "What is happening with Anthropic today, and which coding harness leads SWE-bench?")]
})
print(result["messages"][-1].content)

RAG over AI news

from langchain_community.vectorstores import FAISS
from langchain_openai import OpenAIEmbeddings
from tensorfeed.langchain import TensorFeedNewsLoader

docs = TensorFeedNewsLoader(category="research", limit=100).load()
store = FAISS.from_documents(docs, OpenAIEmbeddings())
retriever = store.as_retriever()

Tool inventory

• tensorfeed_news: latest AI news with category filter
• tensorfeed_status: live AI service status
• tensorfeed_routing_preview: top-1 model recommendation per task
• tensorfeed_attention: AI Attention Index leaderboard
• tensorfeed_harnesses: cross-harness coding-agent leaderboard
• TensorFeedNewsLoader: Document loader for RAG

LlamaIndex

BaseReader implementations for ingesting TensorFeed data into a LlamaIndex VectorStoreIndex or KeywordIndex.

Install

pip install 'tensorfeed[llamaindex]'

Index recent AI news

from llama_index.core import VectorStoreIndex
from tensorfeed.llamaindex import TensorFeedNewsReader

docs = TensorFeedNewsReader(category="research", limit=100).load_data()
index = VectorStoreIndex.from_documents(docs)
qe = index.as_query_engine()
print(qe.query("Summarize this week's most cited research papers"))

Index the AI Attention snapshot

from tensorfeed.llamaindex import TensorFeedAttentionReader

docs = TensorFeedAttentionReader().load_data()
# Each Document is one provider with score, raw signal counts, and recent articles in metadata.
for d in docs:
    print(d.metadata["provider"], d.metadata["attention_score"])

Reader inventory

• TensorFeedNewsReader: AI news as Documents (category, limit)
• TensorFeedAttentionReader: AI Attention Index as one Document per provider

CrewAI

BaseTool implementations for use in any CrewAI Agent. Each tool is a thin wrapper around a free TensorFeed endpoint and returns a string formatted for an LLM to read.

Install

pip install 'tensorfeed[crewai]'

Wire into an Agent

from crewai import Agent, Task, Crew
from tensorfeed.crewai import tensorfeed_tools

researcher = Agent(
    role="AI ecosystem researcher",
    goal="Track frontier AI provider activity and surface real signals.",
    backstory="Expert in real-time AI news, harness benchmarks, and provider status.",
    tools=tensorfeed_tools(),
)

task = Task(
    description="Identify the top 3 AI providers by attention this week and summarize what is driving each.",
    agent=researcher,
    expected_output="A markdown bullet list with provider, score, and the headline driving the score.",
)

Crew(agents=[researcher], tasks=[task]).kickoff()

Tool inventory

• tensorfeed_news_tool: AI news with category and limit
• tensorfeed_status_tool: live AI service status
• tensorfeed_attention_tool: AI Attention Index leaderboard
• tensorfeed_harnesses_tool: cross-harness coding-agent leaderboard
• tensorfeed_routing_tool: top-1 routing recommendation per task