Skip to content

README.md

Latest commit

 

History

History
107 lines (87 loc) · 4.73 KB

README.md

File metadata and controls

107 lines (87 loc) · 4.73 KB

AI-Agent-Lab: Learn Level 1 to Level 5 AI Agents

Practical AI Agents

A collection of simple and advanced AI agents. Each agent is self-contained with a clear entrypoint and local dependencies when needed.

Levels of AI Agents

Level 1: Reactive Agents (Stateless Tools)

  • What they do: Respond directly to prompts/inputs without memory.
  • Examples: ChatGPT answers a question, then translates it into another language
  • Why it matters: Baseline capability — pure input → output.

Level 2: Memory-Enhanced Agents (Contextual Helpers)

  • What they do: Use short-term or long-term memory to improve responses over time.
  • Examples: LangChain agents with vector databases; AI customer support that remembers past chats.
  • Why it matters: Moves from "stateless tool" to "personalized assistant."

Level 3: Tool-Using Agents (API/Environment Operators)

  • What they do: Call APIs, browse the web, interact with databases, or control external apps.
  • Examples: Agents that can use search tools to retrieve time sensitive information
  • Why it matters: Expands beyond conversation into action-taking.

Level 4: Multi-Agent Systems (Collaborative Agents)

  • What they do: Multiple agents with specialized roles work together.
  • Examples:
    • One agent researches, another summarizes, another critiques, working together to generate a final plan (Trip planner)
  • Why it matters: Specialization + collaboration → more robust results.

Level 5: Autonomous Task Agents (Goal-Driven Executors)

  • What they do: Given a goal, they break it into subtasks, plan execution, and complete it with minimal human input.
  • Examples:
    • Workforce that drafts a plan, assign corresponding workers that have different toolkits, then summarizes the final result
  • Why it matters: Represents the highest current level of autonomy that's practical and reproducible today.

Featured Frameworks

  • Python: Primary language for all agents
  • Per-agent dependencies: Declared via local requirements.txt files where applicable

Simple Agents

  • QA-with-search: Answers questions using a lightweight search/retrieval step.
  • QA-with-translate: Answers questions with simple translation for multilingual I/O.
  • Meta-Agent: Workflow Parameter Comparison: Generates examples, runs workflow parameter sweeps, and analyzes differences (see Level 2/comparison-agent/README.md).

Advanced Agents

  • CodeGen-Agent: Program synthesis and evaluation workflows (HumanEval tooling and execution pipeline).
  • Finance-Agent: Multi-step coordination for finance-related tasks (planner/coordinator prompts included).
  • Keypoint-Organizer-Agent: Orchestrates a workforce-style pipeline to extract and organize key points.
  • Trip-Planner-Agent: Multi-step itinerary and trip planning with constraints.

Prerequisites

  • Python: 3.10+
  • Virtual environment: venv or conda (recommended per agent)
  • API keys (if needed): Some agents may require external API keys; check agent code/README

Installation

  1. Clone the repository
git clone https://github.com/your-org/agent-git-repo.git
cd agent-git-repo
  1. Create and activate a virtual environment (recommended)
python3 -m venv .venv
source .venv/bin/activate  # macOS/Linux
# On Windows (PowerShell)
# .venv\\Scripts\\Activate.ps1
  1. Install dependencies per agent (install only what you need)
# Finance-Agent
echo "Installing Finance-Agent deps"
pip install -r Advanced-Agents/Finance-Agent/requirements.txt

# Keypoint-Organizer-Agent
echo "Installing Keypoint-Organizer-Agent deps"
pip install -r Advanced-Agents/Keypoint-Organizer-Agent/requirements.txt

# Trip-Planner-Agent
echo "Installing Trip-Planner-Agent deps"
pip install -r Advanced-Agents/Trip-Planner-Agent/requirements.txt
  1. Run an agent
# Simple agents
python Simple-Agents/QA-with-search/qa_with_search.py
python Simple-Agents/QA-with-translate/QA-with-Trans.py

# Advanced agents (examples)
python Advanced-Agents/Finance-Agent/finance_agent.py
python Advanced-Agents/Trip-Planner-Agent/trip_planner.py
python Advanced-Agents/Keypoint-Organizer-Agent/keypoint_workforce.py
# CodeGen-Agent entrypoints
python Advanced-Agents/CodeGen-Agent/workflow.py
python Advanced-Agents/CodeGen-Agent/humaneval/humaneval.py

Contributing

  • Issues: Report bugs and request features via issues
  • Branches/PRs: Use feature branches; open PRs with clear descriptions
  • Style: Favor clear, readable Python; keep dependencies scoped per agent
  • Docs: Update this README or add agent-level READMEs for new agents
  • Tests/Examples: Include minimal tests or usage examples where applicable

Thanks for contributing and exploring these agents!