Agent architecture

This page is the index for agent-architecture articles.

It focuses on engineering concerns in multi-step, tool-using (agentic) systems.

Start here (pick the job)

LLM Memory Boundary Model: Context Construction (Eligibility, Selection, Persistence) and Why Answers Change
A concrete model for what enters context and why (selection + persistence)
Next: Manage LLM memory boundaries (ChatGPT + agentic systems) — procedure
LLM-Led vs Orchestrator-Led Tool Execution Control-Plane Placement Tradeoffs
A decision on where orchestration and policy enforcement live (model-led vs orchestrator-led)
Next: The attack surface is the orchestration loop, not the model
Human vs GenAI capability map (engineering view)
A framing for what should be automated vs kept human-owned (capability + failure modes)
Next: Engineering Quality Gate — choose the correct procedure

Pages in this section

Core pages for this topic are listed above under “Start here”. Additional pages will appear here as they’re published.

Section resources

About this section About this section

Focus

  • Orchestration patterns (control-flow mechanisms) — where control flow lives and how workflows are sequenced.
  • State & lifecycle management — what persists across steps and boundaries (session/thread/run), what resets, and when.
  • Tool invocation lifecycle — selection, authorization/enforcement, validation, error handling, retries, and egress constraints.
  • Write paths — where and how the system can persist changes to external systems.

Terminology

  • Orchestrator: the component that owns control flow (decides next action/tool call) and enforces policy.
  • Model-led: the model proposes next actions/tool calls within constraints enforced by the orchestrator.
Related Related
External baselines (shared terminology) External baselines (shared terminology)