Ai Agents Guides

A four-layer walkthrough for running Claude Code without babysitting it: auto mode to cut approval prompts, /goal to enforce a finish line, and /schedule Routines for cloud-scheduled jobs. Every feature claim from a popular June 2026 video is re-checked against Anthropic docs and corrected where the video got it wrong — /routines is really /schedule, and Claude for Chrome is no longer macOS-only.

Level: beginner · 15 min read

The five agentic patterns from Anthropic's 'Building Effective Agents' — prompt chaining, routing, parallelization, orchestrator-workers, and evaluator-optimizer — each with one runnable Python example. Start simple, add complexity only when it pays for itself.

Level: intermediate · 8 min read

A small local model that runs on your own desk can do real work — but only if you stop prompting it like ChatGPT. This guide breaks down intelligence per token and the five-step agentic loop that turns a 3B-active model into a reliable builder, with the exact specify-implement-review-patch-verify cycle and the hardware and tools to run it.

Level: intermediate · 22 min read

A practical decision framework for choosing between a persistent background AI agent (a claw) and a prompt-triggered on-demand model. Includes a decision tree with four checkpoints mapped to real enterprise scenarios.

Level: intermediate · 27 min read

Build a Kanban State Machine for AI Agents in Plain Node.js When you run more than one AI agent at a time, a chat log stops being enough. You need to know — at a glance — what is running, what is waiting on you, and what is queued.

Level: intermediate · 10 min read

A small Anthropic-popularized prompting trick: instead of asking your AI agent to reply in plain text, ask it for HTML. You get scannable tables, color-coded options, and diagrams instead of a wall of prose. This guide covers the exact prompt, a before/after readability comparison, the real use cases, and the honest tradeoffs (token cost, noisy diffs) so you know when NOT to use it.

Level: beginner · 11 min read

GitHub instrumented its own agentic workflows and found agents quietly burning tokens on overhead nobody asked for. This guide breaks down the three fixes that cut real cost by 19–62%, and how to measure efficiency instead of raw volume.

Level: intermediate · 11 min read

Gemini Skills: The Google Feature Saving Teams Hours (Almost No One Knows About It) At Google Cloud Next '26 in April, Google rolled out a stack of agentic Workspace features.

Level: beginner · 20 min read

Stop coupling your agent to a single model. Build a durable work loop, keep memory outside the brain, and route each step to the model that fits.

Level: intermediate · 18 min read

A deep look at run_script — the AWS MCP Server tool that lets agents chain multiple AWS API calls in one server-side Python execution, with the IAM, network, and debugging tradeoffs developers should understand before turning it on in production.

Level: intermediate · 15 min read

A technical 2026 comparison of Cursor 3's Agents Window and Claude Code across autonomy, planning, parallelism, verification, and team fit. Verdict: Cursor wins the visual multi-agent workflow; Claude Code keeps the agentic edge on deeper autonomous execution.

Level: intermediate · 17 min read