Skip to content

secrethash/agents-in-python

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

3 Commits
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Agents in Python

This is a personal learning path repository for learning to build agents in python.

NOTE: This is not meant to teach, but to learn.

👩‍💻 Learning Path

The learning path is divided into 3 Tiers:


🟢 Tier 1 — Fundamentals (Tool Use, Memory, Planning)

An agent that can use Python functions as tools — a calculator, a weather fetcher, a file reader. This is the "Hello World" of agents. Learn the ReAct loop (Reason → Act → Observe) which is the heartbeat of every agent.

An agent that remembers things across conversations using a simple vector store (like ChromaDB). Understand the difference between short-term context and long-term memory — a key design decision in every real agent.

Give the agent a high-level goal (e.g., "Research Python best practices and write a summary") and it breaks it into subtasks, executes them sequentially, and reports back. Core concept: planning and self-prompting.


🟡 Tier 2 — Real-World Automation

An agent that takes a question, searches the web (via Tavily/SerpAPI), scrapes relevant pages, synthesizes an answer, and cites sources. Combines tool use, chaining, and output formatting.

Point the agent at a local repo and ask it questions: "Find all functions that handle auth", "Refactor this module". You'll learn RAG + agents together with file I/O tools.

Connect to Gmail and Google Calendar APIs. The agent reads emails, drafts replies, schedules meetings, and handles follow-ups. This is where real-world reliability and error handling become the main lesson.


🔴 Tier 3 — Multi-Agent Systems

Two agents with distinct roles: a Researcher agent that gathers data, and a Writer agent that turns it into a blog post or report. They communicate via a shared message bus. This teaches agent orchestration.

Give a spec, and a team of agents (Planner → Coder → Tester → Debugger) collaboratively write, run, test, and fix code. This is the capstone — it ties together everything from all tiers.


📍 Recommended Learning Stack

Layer Recommended Used
LLM API Anthropic (claude-sonnet) or OpenAI Ollama + qwen2.5:7B
Agent Framework Raw Python first, then LangGraph Raw Python first, then LangGraph
Memory ChromaDB or FAISS ChromaDB
Tool Execution Function calling / MCP Function calling
Orchestration LangGraph or AutoGen (Tier 3)

About

This is a personal learning repository for Agents in Python

Resources

Stars

Watchers

Forks

Contributors

Languages