In this repository, I'm going to explain the fundamentals of agent development using general-purpose programming languages and real-world examples based on my own projects. My goal is to create easy-to-follow learning content and videos based on practical examples, covering a complete series on agent development. The goal is to help programmers understand how to build agentic systems and make the right engineering trade-offs. Agentic development is still a relatively new field, but many useful articles and resources have already been published. Throughout this repository, I'll try to reference relevant resources and research as I build and explain each topic.
- How do LLMs work?
- What is an Agent?
- What is State?
- What is Memory?
- What is Knowledge?
- What is a Chunk?
- What is the Agent Lifecycle?
- How does Tool Calling work?
- What is MCP and how does it work?
- Tool Calling vs. MCP
- Tool Design & Schemas
- MCP Discovery
- What is Agent Architecture?
- Agent State
- Agent Context
- Agent Memory
- Agent Orchestration
- Single-Agent vs. Multi-Agent
- Common Agent Architectures
- What is Context Engineering?
- Context Window
- Context Selection
- Context Compression
- Context Isolation
- Context vs. Memory vs. Knowledge
- What is RAG?
- Chunking Strategies
- Query Transformations
- Retrieval & Hybrid Search
- Reranking & Compression
- Agentic RAG
- RAG Trade-offs & Optimization
- Short-term vs. Long-term Memory
- Semantic Memory
- Episodic Memory
- Procedural Memory
- Reflection & Experience
- Human Feedback & Memory
- Agent Failure Modes
- Retry
- Idempotency
- Timeout
- Caching
- Agent Failure Recovery
- Deterministic vs. Probabilistic Behavior
- Human-in-the-Loop (HITL)
- Why Multiple Agents?
- Agent Delegation
- Agent Communication
- Supervisor Pattern
- Specialist Agents
- Parallel Agents
- Multi-Agent Workflows
- Trade-offs of Multi-Agent Systems
- Why is Agent Evaluation Different?
- Step-level Evaluation
- Final Evaluation
- Tool-call Evaluation
- Trajectory Evaluation
- LLM-as-a-Judge
- Evaluation Datasets
- Human Evaluation
- Regression Testing
- Agent Observability
- Tracing
- Logging
- Metrics
- Token Monitoring
- Cost Monitoring
- Agent Performance Monitoring
- Prompt Injection
- Tool Permissions
- Agent Authorization
- Data Leakage
- Untrusted Tool Output
- Guardrails
- Input Validation
- Output Validation
- Agent Governance
- Permissions
- Policies
- Auditability
- Human Oversight
- Compliance
- Safe Agent Execution
- Token Optimization
- Context Optimization
- Model Selection
- Caching Strategies
- Parallel Tool Calling
- Concurrency
- Latency Optimization
- Cost vs. Quality Trade-offs
- Designing Production-ready Agents
- Agent Architecture Trade-offs
- Failure Recovery
- Scalability
- Reliability
- Observability
- Security
- Evaluation
- Cost Optimization
- Continuous Improvement