Explainer
Every explainer piece on DeskTheory, newest first.
-
When to use /goal vs. /loop in Claude Code
Two Claude Code commands that both keep it running without you re-prompting. `/goal` drives one job forward, turn after turn, until a finish line you define is provably met. `/loop` re-runs a prompt…
-
What is Claude Mythos?
Claude Mythos is Anthropic's vulnerability-finding model: a general-purpose AI that hunts security holes in real software faster than any human team. Anthropic announced it on April 7, 2026, and is…
-
What is the Compound Engineering plugin for Claude Code?
The Compound Engineering plugin is a free add-on for [Claude Code](/learn/claude-code) that bolts a plan, build, review, learn loop plus a library of 37 skills and 51 agents onto your terminal. It's…
-
Claude Code in the terminal vs the desktop app: which should you use?
Same engine, two front doors. The terminal is the scriptable command line you type into; the desktop app is that same agent in a windowed workspace with visual diffs and parallel sessions side by…
-
When to use a goal vs. a dynamic workflow in Claude Code
Two new Claude Code features that both let it keep working without you. A dynamic workflow goes wide. A goal keeps going. One question tells you which to reach for.
-
How do I keep my OpenCLAW agent from posting things on the internet?
Short version: your agent posts and sends with your credentials, on the channels you connected, so you keep it from going rogue with three things. A human-approval gate on every outbound action, a…
-
What are dynamic workflows in Claude Code?
Claude Code's new ability to plan a big job, run tens to hundreds of agents on its pieces in parallel, check the work, and hand you one finished result. Anthropic shipped it on May 28, 2026.
-
Memory and Projects for beginners: what they are and when to switch them on
Every chat you open starts with no memory of you, and once you understand why, the two fixes take about ten minutes to set up.
-
Spotting a confidently wrong answer before it burns you
A lawyer filed a brief packed with real-looking case citations, all invented by AI, and didn't catch a single one before it landed in court.
-
What is Microsoft Scout and how can you use it?
Microsoft's first always-on "Autopilot" agent. It runs on the exact open [harness](/articles/what-is-a-harness) operators here already build with, [OpenCLAW](/articles/what-is-openclaw). The…
-
When not to use AI: the work to keep human
A lawyer submitted a brief with six case citations. Every case was invented by AI. The citations looked completely real. The lawyer paid the price.
-
Desktop, browser, or phone: where should a CEO use AI?
The surface is just the window. Your account is the home.
-
What AI can and cannot do for a CEO right now
In a Harvard and BCG experiment with 758 consultants, AI made them about 40% better at some tasks and about 19% worse at others. The tasks looked identical. That gap is the whole game.
-
How AI actually works (plain English)
Under the hood of every chat tab: a pattern machine that predicts the next word, one token at a time. Beautiful fluency, no built-in truth filter, and a working memory that erases itself when the…
-
What is a prompt?
Everything you type into an AI chat is a prompt. But a prompt is more than a question. It's the full package you hand the model: task, context, format, and constraints. Most first prompts deliver…
-
Is your data safe in AI?
The honest answer depends on one thing almost no one checks: which plan you are on. Consumer plans and business plans treat your data completely differently, and most leaks happen because someone…
-
What are connectors in Claude Code?
The plug that lets Claude reach into the tools you already use, your email, Slack, your numbers, and do the work there instead of just talking about it.
-
What is a context window?
The model's working memory: everything it can see right now, measured in tokens. If the thing it needs isn't in the window, the model doesn't ask for it. It guesses.
-
What is a frontier model?
The handful of most capable AI models at the leading edge, the ones that cost hundreds of millions of dollars to train and reset the bar every few months. When the news says "the frontier," this is…
-
What is adverse impact?
The legal idea that a hiring practice can be illegal even when you had zero intent to discriminate, if it screens out a protected group at a meaningfully higher rate. It is the single biggest reason…
-
What is an AI agent?
An AI model put to work in a loop: it decides, takes an action, looks at the result, and repeats until the job is done. A chatbot talks and stops. An agent acts until the work is finished.
-
What is an ATS?
Applicant tracking system. The software where every job application lands, gets sorted, and moves through your hiring pipeline. If you have ever posted a role and watched 300 resumes pile up…
-
What is an HRIS?
Human resources information system. The system of record for your employees: payroll, benefits, time off, onboarding, the org chart. Everything about a person from the day they accept the offer to…
-
What is a large language model (LLM)?
The engine inside ChatGPT, Claude, and Gemini. It predicts the next word from patterns it read across the internet. Brilliant at fluency, indifferent to truth, and that one fact is most of what a…
-
What is a Claude Project?
A saved workspace inside Claude that already knows your company. You load it once with your context and your instructions, and every chat you start inside it begins smart instead of from zero.
-
What is gbrain?
A memory layer for your AI agent, open-sourced by the CEO of Y Combinator. It is the fix for the thing that quietly breaks every agent: it forgets.
-
What is Google Antigravity?
Google's agent harness. The engine room underneath Gemini Spark, and Google's answer to the open harness layer that OpenCLAW and Hermes occupy.
-
What is Google Spark and how can you use it?
Google's 24/7 personal AI agent. The mainstream, no-terminal version of the "agent that works while you sleep" that operators here build with Claude Code and OpenCLAW.
-
What is MCP?
The Model Context Protocol. The universal plug that lets any AI app connect to any tool or data source. Think USB-C for agents: one port, and everything that speaks it just works.
-
What is RAG?
Retrieval-augmented generation. The trick that turns an AI that knows the internet into an AI that knows your company. Every "chat with your docs" tool you have ever touched runs on it.
-
What is shadow AI?
Your team is already using AI for work. Just not the AI you approved, on accounts you cannot see, with data you would never have signed off on. That gap between the AI you sanctioned and the AI they…
-
What is the AskUserQuestion tool?
The built-in tool Claude Code uses to stop mid-task and ask you a quick multiple-choice question instead of guessing. The closest thing the terminal has to a chief of staff who checks in before…
-
What is vibe coding?
Describing software in plain English and letting an AI write it, without ever reading the code. The fastest way a non-technical CEO can get a working tool. Also the fastest way to ship something…
-
What is a coding agent?
The category that Claude Code, Codex, and Cursor all belong to. Not an AI that talks about code. One that reads your files, edits them, runs them, and checks its own work.
-
What is agentskills.io?
The open standard that makes the skills you build portable. Write a capability once, and run it on whatever tool you use next.
-
What is Codex?
OpenAI's coding agent. The same idea as Claude Code, wearing an OpenAI badge, and probably already bundled in the ChatGPT plan you pay for every month.
-
What is a cloud VM?
A computer you rent that lives in someone else's data center. Always on. Accessible from anywhere. Where you run an agent when you do not want to keep your laptop open.
-
What is a markdown file?
A text file with a tiny bit of formatting. The format every AI tool reads natively, which is why my whole AI stack is built on it.
-
What is a SQLite database?
A whole database that lives in a single file on your computer. No server. No admin. No login. The thing most agent [harnesses](/articles/what-is-a-harness) use to give your agent a memory.
-
What are skills in Claude Code?
The folder that turns a workflow you ran once into a capability Claude executes on command, forever.
-
What is a harness?
The layer that allows your language model to get real work done. The thing that turns a chatbot into an operator that gets smarter every day.
-
What is a slash command?
A prompt you save once and invoke from any Claude Code session by typing `/name`. The lightest-weight way to ship a skill.
-
What is Granola?
The meeting recorder that turns every conversation you have into searchable [markdown](/articles/what-is-a-markdown-file) your AI can read.
-
What is a CLAUDE.md file?
The markdown file Claude reads on startup so it stops asking "what business are we in?" every time you open it.
-
What is a cron?
The thing that makes your agent wake up on its own and do things for you without you having to be there to ask.
-
What is Hermes?
The open-source agent harness from Nous Research. Mobile-first. Multi-model. The main alternative to the OpenCLAW harness I run.
-
What is OpenCLAW?
It’s a [harness][8] (I’ll explain). In short, it’s the thing that turns an LLM into an operator that gets sharper every day.\ It is 6:47 AM on a Tuesday. My OpenCLAW agent has already done the…