Roadmap¶

Published¶

The book is available on Amazon. Thirteen chapters across four parts covering the full lifecycle of building production agent systems -- from first principles through governance, security, memory, and protocols.

This site is the code companion -- working implementations, tests, diagrams, and evaluation evidence.

Foundations (free)¶

Five hands-on sections published as a free pre-read for readers new to LLMs and agentic AI:

• How LLMs Actually Work -- APIs, tokens, context windows, hallucination
• From API Calls to Tool Use -- Function calling, schema validation, tool execution
• Your First Agent, No Framework -- Complete agent in 100 lines, end-to-end
• The Same Agent, With a Framework -- ADK and LangChain side-by-side with eval comparison
• Connecting Your Agent to MCP -- Build an MCP server, connect to real tools and services

Shipped in this repo¶

• Working code for every concept: tool registry, context pipeline, agent loop, workflow implementation, bounded agent, state management, multi-agent orchestration, approval gates, escalation engine, audit logging, eval harness, tracer, reliability hardening, cost profiler, security hardening, session memory, long-term memory, shared memory, memory security
• 3 end-to-end projects: Document Intelligence Agent, Incident Runbook Agent, and Memory Agent
• Eval harness with gold dataset, rubric, scored comparison script, and failure buckets
• 130+ passing tests across unit and integration suites
• 40+ architecture-grade diagrams (hand-crafted SVGs)
• Infrastructure: pyproject.toml, Makefile, .env.example

What might come next¶

• Code examples for Chapters 8-11 (metacognition, deployment, governance, security)
• Code examples for Chapter 13 (MCP server patterns, AIP delegation chains)
• Additional eval datasets and adversarial test suites
• Community contributions: real-world case studies, production deployment patterns

Content ships when it meets the quality bar. No timelines promised.