---
title: "Building blocks"
id: "building-blocks"
pass: "Pass 1: Building Blocks"
canonical_url: "https://renatomignone.github.io/From-LLMs-to-Secure-Agents/building-blocks/"
total_chapters: 10
---

# Building blocks

> **Section Overview:** Provide the complete functional anatomy of production agentic systems.
> **Pass:** Pass 1: Building Blocks · **Published chapters:** 10

## Learning Outcomes
The reader can locate every model call, context item, plan, state transition, memory operation, retrieval step, tool action, identity, execution boundary, human control, trace, evaluation, agent message, adaptation step, operational mechanism, and artifact.

## Published chapters
- [Model roles and selection](https://renatomignone.github.io/From-LLMs-to-Secure-Agents/building-blocks/01-model-roles-and-selection/) - Explains model roles, capability profiles, selection dimensions, provider adapters, and cost-latency-quality trade-offs in production agentic systems.
- [Routing, cascades, and fallbacks](https://renatomignone.github.io/From-LLMs-to-Secure-Agents/building-blocks/02-routing-cascades-and-fallbacks/) - Explores dynamic model routing, progressive escalation cascades, circuit breaker patterns, and multi-provider fallbacks for high-availability agent architectures.
- [Capability, cost, latency, and reliability](https://renatomignone.github.io/From-LLMs-to-Secure-Agents/building-blocks/03-capability-cost-latency-and-reliability/) - Analyzes the multi-dimensional trade-offs between model capability, token economics, latency profiles, and operational reliability in production agent workflows.
- [Routing evaluation](https://renatomignone.github.io/From-LLMs-to-Secure-Agents/building-blocks/04-routing-evaluation/) - Explains quantitative evaluation frameworks, cost-quality Pareto benchmarks, and judge calibration techniques for model routers and cascades.
- [Context sources and precedence](https://renatomignone.github.io/From-LLMs-to-Secure-Agents/building-blocks/01-context-sources-and-precedence/) - Explains the runtime discipline of context engineering, classifying dynamic prompt sources, trust boundaries, and precedence hierarchies in agent loops.
- [Context budgets, selection, and ordering](https://renatomignone.github.io/From-LLMs-to-Secure-Agents/building-blocks/02-context-budgets-selection-and-ordering/) - Explains quantitative token budgeting, knapsack chunk selection algorithms, and attention-anchored positional ordering in agent context windows.
- [History, summaries, and compression](https://renatomignone.github.io/From-LLMs-to-Secure-Agents/building-blocks/03-history-summaries-and-compression/) - Explains conversation history management, lossy and lossless token compaction, tool trace pruning, and recursive summarization for long-running agents.
- [Provenance and context debugging](https://renatomignone.github.io/From-LLMs-to-Secure-Agents/building-blocks/04-provenance-and-context-debugging/) - Explains token-level lineage tracking, context debugging inspection consoles, OpenTelemetry telemetry standards, and boundary integrity auditing.
- [Reactive and reason-act patterns](https://renatomignone.github.io/From-LLMs-to-Secure-Agents/building-blocks/01-reactive-and-reason-act-patterns/) - Explains reactive action selection and interleaved ReAct (Thought-Action-Observation) loops, error recovery, and cycle bounds in autonomous systems.
- [Decomposition and plan-execute](https://renatomignone.github.io/From-LLMs-to-Secure-Agents/building-blocks/02-decomposition-and-plan-execute/) - Explains multi-step goal decomposition, the two-tier Planner-Executor pattern, subtask lifecycle management, and dynamic replanning protocols.