THE EDITORIAL ARCHIVE
Perspectives on technology, certifications and practical IT skills.
Build permission-aware Azure AI Search RAG with hybrid retrieval, vector embeddings, source provenance, relevance tests, and a clear classic-versus-agentic decision.
Read articleBuild defensible AI-103 multi-agent workflows with scoped specialists, typed handoffs, durable state, partial failure recovery, and measured orchestration value.
Read articleA practical Microsoft Foundry architecture for trustworthy AI agents, with secure retrieval, bounded tool contracts, approval state, source provenance, and failure-aware release engineering.
Read articleDesign evidence-led multi-agent Claude research: coordinator, searcher, analyst and writer contracts, source ledgers, uncertainty and reproducible reports.
Read articleUnderstand why Exam-Topics lists CCA-F while Anthropic's current Architect Foundations exam guide uses CCAR-F; distinguish Associate, Developer and Professional.
Read articleDecompose Claude agent tasks into safe, evidence-led workflows and select built-in, client and server tools by authority and execution boundary.
Read articleUse Claude Code path-scoped rules and a verified exploration trail to refine changes without losing context, test evidence, or permissions.
Read articleSet explicit Claude criteria, validate semantics, propagate partial agent failures and calibrate human review using observed risks.
Read articleDesign a testable Claude support agent with tools, safe delegation, policy gates, typed errors, structured evidence and session recovery.
Read articleHow Claude Agent SDK tool loops, stop reasons, hooks, bounded retries and safe execution boundaries fit together.
Read articleDesign dependable Claude subagent delegation, handoffs, provenance and recoverable sessions.
Read articleUse Claude Code plan mode, skills, hooks, permission boundaries and CI workflows with reliable review controls.
Read articleLearn when to use Claude few-shot examples, Message Batches and multi-pass evidence review.
Read articleImprove Claude MCP tool choice, descriptions and structured error recovery with bounded permissions.
Read articleCompare direct Claude API access, Claude Platform on AWS, Amazon Bedrock and Vertex AI across control planes, model choice, governance and recovery.
Read article