The lab notebook
Field notes on AI engineering
824 articles · page 14 of 28
Agent Frameworks for Beginners: Where to Start in 2026
A practical, no-hype guide to picking your first agent framework in 2026 and actually shipping something with it.
Agent Guardrails: Preventing Runaway Loops and Cost Overruns
Learn how to cap iterations, budget tokens, and detect loops so autonomous agents fail safely instead of racking up runaway costs.
Agent Handoff Patterns: When One Agent Passes Work to Another
A practical guide to designing clean handoffs between AI agents, with code patterns for context passing, state validation, and failure recovery.
Agent Memory Architectures Compared: Buffer, Vector, Graph
A practical comparison of buffer, vector, and graph memory for AI agents, with code and guidance on when to use each.
Agent Observability: Tracing Every Decision an Agent Makes
Learn how to trace, log, and debug autonomous agents with spans, structured events, and decision graphs before they fail in production.
Agent Personas: Designing Consistent Behavior and Tone
Learn how to design AI agent personas that stay consistent across sessions, tools, and edge cases instead of drifting into generic chatbot tone.
Agent Prompt Injection: Attack Vectors and Defenses
A practical breakdown of how prompt injection attacks compromise AI agents, with real attack patterns and layered defenses you can ship today.
Agent Retry Logic: Handling Transient Tool Failures Gracefully
A practical guide to designing exponential backoff, retry budgets, and circuit breakers so AI agents survive flaky tools instead of collapsing on the first error.
Agent SLAs: Setting Realistic Uptime and Latency Expectations
How to define agent SLAs that survive contact with LLM variance, tool failures, and real production traffic.
Agent State Management: Where Does Agent State Actually Live?
A practical breakdown of where AI agent state really lives, from context windows to databases, and how to design agent state management that survives crashes.
Agent Team Structures: Who Owns What in a Multi-Agent System
A practical guide to designing ownership boundaries in multi-agent systems, so agents coordinate instead of colliding.
Agent Testing Strategies: Unit, Integration and End-to-End
A practical guide to testing AI agents at the unit, integration, and end-to-end level, with code you can actually run.
The Skills Gap Between Agent Tutorials and Real Agent Jobs
Tutorials teach you to call a tool. Real agent jobs need error handling, observability, cost control, and evals — here's the gap and how to close it.
Agent Version Control: Managing Prompt and Tool Changes Over Time
A practical guide to versioning prompts, tools, and configs for AI agents so changes stop breaking production silently.
Portfolio Projects That Get You Hired as an AI Agent Engineer
Six concrete, scoped AI agent projects — with architecture and code — that prove real engineering skill to hiring managers.
Building a Coding Agent from Scratch: Core Components Explained
A practical breakdown of the core components behind coding agents like Claude Code, with working Python examples for each piece.
Building a Custom LLM Judge Rubric From Scratch
A practical, code-first guide to designing, testing, and hardening a custom LLM judge rubric instead of reaching for generic "rate 1-10" prompts.
Building an Eval Dashboard: Tracking Quality Over Time
Learn how to design and build an LLM eval dashboard that tracks quality over time, catches regressions early, and turns vibes-based debugging into real data.
Building an Eval Team: Roles and Responsibilities
A practical breakdown of the roles a real LLM eval team needs, from data curation to eval engineering, and how they work together.
Building a Golden Dataset for LLM Evaluation
A practical guide to designing, labeling, and maintaining the golden dataset that makes your LLM evals trustworthy.
Building in Public: Sharing Your Agent Bootcamp Progress
A practical guide to documenting your AI agent bootcamp journey publicly, turning daily learning into a portfolio, a network, and a job pipeline.
Choosing Between Codex, Claude Code and Copilot for Your Team
A practical, no-hype comparison of OpenAI Codex, Claude Code, and GitHub Copilot to help engineering teams pick the right AI coding agent.
Claude Code for API Development: Design to Implementation
Learn how Claude Code turns API design into working, tested endpoints using a practical, repeatable engineering workflow.
Claude Code Context Window Management: Avoiding Context Rot
Learn how Claude Code's context window works and the concrete habits — /clear, /compact, subagents — that stop long sessions from degrading into context rot.
Claude Code Cost Management: Understanding Token Usage
Learn how Claude Code spends tokens, why context windows drain fast, and practical habits that cut your bill without cutting output quality.
Claude Code Custom Slash Commands: Building Your Own Shortcuts
Learn how to build custom Claude Code slash commands with real .claude/commands examples for reviews, tests, commits, and more.
Claude Code for Data Pipelines: ETL Development Workflow
A practical workflow for building, debugging, and testing ETL pipelines with Claude Code, from schema exploration to production-ready transforms.
Claude Code for Database Migrations: A Safe Workflow
A concrete, safe workflow for using Claude Code to write, review, and roll back database migrations without breaking production.
Claude Code Extended Thinking: When to Ask It to Think Harder
Learn when "think," "think hard," and "ultrathink" actually change Claude Code's output, and when they just burn tokens for nothing.
Claude Code for Accessibility: Auditing and Fixing A11y Issues
Learn how to use Claude Code to audit, diagnose, and fix real accessibility issues with concrete ARIA and semantic HTML examples.