Agentic Systems
Systems that choose their own next step. Architecture patterns, tool design and MCP, context engineering, trajectory evaluation, and the operational reality of software that acts on its own.
Know when autonomy earns its unpredictability — and when it does not.
Practitioner
3Agent Architecture Patterns
Single loop, prompt chain, router, orchestrator-worker, parallel fan-out and multi-agent. What each pattern is genuinely for, the coordination cost that multi-agent quietly reintroduces, and a selection table you can argue from.
9 min readagentsarchitectureEvaluating Agents
The unit of evaluation is a trajectory, not a response. Outcome versus trajectory assertions, step and cost budgets as first-class checks, replay harnesses, and why an agent that succeeds expensively has still failed.
9 min readevaluationtestingTool Design And The Model Context Protocol
The quality of an agent is mostly the quality of its tools. How to design interfaces for a model rather than a human, why error messages are instructions, and what MCP standardises without solving.
10 min readtool usemcp
Advanced
3Context Engineering And Agent Memory
Context is a scarce, expensive, actively managed resource rather than a place to put everything. Working versus persistent memory, what compaction loses, sub-agent isolation, and why long-running agents degrade.
10 min readcontext engineeringmemoryDesigning Agents With Bounded Authority
The design question for an agent is not which model it uses but what it is permitted to do. Authority boundaries, idempotency, blast-radius limits, prompt injection as a live threat, and audit trails that hold up.
10 min readagentstool useRunning Agents In Production
The operational surface of a system that acts on its own. Sandboxing, durable execution, circuit breakers and runaway cost, rate limiting, trajectory observability, on-call, and how you stop one mid-run.
10 min readoperationssandboxing
Other domains