Context-driven agents that work out of the box.
from cogency import Agent
agent = Agent()
result = await agent("Search for Python best practices and summarize")Zero ceremony. Maximum capability.
- 🌐 Web-enabled - Search and scrape with zero configuration
- 🔌 Multi-provider - OpenAI, Anthropic, Gemini support
- 🛠️ Tool orchestration - Files, shell, web tools auto-compose
- 🧠 Context injection - Automatic assembly of relevant information
- ⚡️ Streaming - Event-coordinated ReAct reasoning
import asyncio
from cogency import Agent
async def main():
agent = Agent()
response = await agent("What are the benefits of async/await in Python?")
print(response)
# Run with: python -m asyncio your_script.py
asyncio.run(main())pip install cogencySet your API key:
export OPENAI_API_KEY="your-api-key-here"from cogency import Agent
agent = Agent()
response = await agent("Explain quantum computing in simple terms")from cogency import Agent, BASIC_TOOLS
from cogency.tools import Search, Scrape
# Web-enabled agent
agent = Agent(tools=[Search(), Scrape()])
result = await agent("Search for Python best practices and summarize key points")
# All basic tools (Files, Shell)
agent = Agent(tools=BASIC_TOOLS)
result = await agent("Create a Python script that calculates factorial of 10")from cogency import Agent, profile
# Set user preferences (optional)
profile("alice",
name="Alice Johnson",
preferences=["Python", "Machine Learning"],
context="Senior data scientist working on NLP projects")
agent = Agent()
response = await agent("Recommend a good ML library for text processing", user_id="alice")from cogency.storage import add_document
# Add documents to knowledge base (optional)
add_document("python_guide", "Python is a high-level programming language...")
add_document("ml_basics", "Machine learning is a subset of artificial intelligence...")
# Agent automatically searches relevant documents for context
agent = Agent()
response = await agent("What's the difference between Python and machine learning?")Context-driven agents work by injecting relevant information before each query:
async def agent_call(query: str, user_id: str = "default") -> str:
ctx = context(query, user_id) # Assembles relevant context
prompt = f"{ctx}\n\nQuery: {query}"
return await llm.generate(prompt)Context sources include:
- System: Base instructions
- Conversation: Recent message history
- Knowledge: Semantic search results
- Memory: User profile and preferences
- Working: Tool execution history
- Zero writes during reasoning - no database operations in the hot path
- Pure functions for context assembly - deterministic and testable
- Read-only context sources - graceful degradation on failures
- Optional persistence - conversation history saved asynchronously
Simple conversational agent with context injection.
agent = Agent()
response = await agent(query: str, user_id: str = "default") -> strfrom cogency import Agent
agent = Agent()
async for event in agent.stream("Complex research task requiring multiple steps"):
if event["type"] == "reasoning":
print(f"Thinking: {event['content'][:100]}...")
elif event["type"] == "complete":
print(f"Final: {event['answer']}")from cogency import Agent
from cogency.lib.providers import Anthropic, Nomic
# Provider agnostic: Mix any LLM with any Embedder
agent = Agent(llm=Anthropic(), embedder=Nomic())
result = await agent("Compare Python and Rust for systems programming")from cogency import profile
from cogency.storage import add_document
# User profiles
profile(user_id, name=None, preferences=None, context=None)
# Knowledge base
add_document(doc_id: str, content: str, metadata: dict = None)# Install dev dependencies
poetry install
# Run tests
pytest tests/See docs/blueprint.md for complete technical specification.
That's it. No configuration, no setup, just working agents.
v2.1.0: Web capabilities, multi-provider support, Result types - zero ceremony preserved.
