Catalog

AI Guides

A catalog of AI tools and techniques — what each one is, when to use it and, above all, when not to.

A new AI tool ships every week, along with a new technique for using it. The hard part is never finding out they exist: it is knowing which one solves your problem today and which one only adds complexity. Every entry here starts with that decision and only then becomes a tutorial. The catalog grows over time — the first entries cover assisted coding, context engineering and multi-agent systems.

TypeTopic

7 entries

Tool · Anthropic

Claude Code

Anthropic's coding agent that lives in the terminal and works inside your repository.

IS: A coding agent that runs in your terminal, reads and edits your project files, runs commands and commits — instead of only suggesting the next line.
Use when: When the task spans several files, requires understanding the whole repository, or is repetitive enough that you can describe it in one sentence.
Avoid when: When you cannot state the expected result, when the code is critical and nobody will review it, or when the change is two lines you would type faster yourself.

Tool · OpenAI

Codex

OpenAI's coding agent, available in your terminal, your IDE, the cloud and inside your pull requests.

IS: An OpenAI coding agent that runs the same task in four places: your terminal, an IDE extension, a cloud sandbox and automatic pull request review.
Use when: When you want to dispatch well-scoped tasks and go do something else — several in parallel — or when you want an automatic first review on every PR.
Avoid when: When the task needs back-and-forth with you at every step, or when nobody on the team has time to review what comes back from the cloud.

Tool · Anysphere

Cursor

The editor that put AI at the centre of code editing — and became the default way to program with assistance.

IS: A code editor — a VS Code fork — where AI is not a plugin but part of how you edit files, navigate and refactor.
Use when: When you want to keep your hands on the wheel: edit, see the diff instantly, adjust and move on, with the least friction between thinking and applying.
Avoid when: When your team is locked to a specific IDE, when code cannot leave the internal network, or when you need automation with nobody watching.

Tool · Google

Antigravity

Google's bet on an IDE where you manage agents instead of writing code.

IS: An agent-first IDE from Google, built on Gemini, where the main screen is a manager of agents working in parallel — not an open file.
Use when: When you want to prototype fast by delegating parallel tasks and want the agent itself to verify the result in a browser before handing it to you.
Avoid when: When the work is production code with real risk, when budget must be predictable, or when the team cannot absorb frequent product changes.

Concept

Context rot

Why filling the context window makes answers worse — and what to do about it.

IS: The measurable drop in LLM quality as the number of input tokens grows — even on simple tasks and even well inside the advertised window.
Use when: Use the concept whenever you decide how much context to inject: in RAG, in long-running agents and in any prompt that grows over time.
Avoid when: It is not an excuse to withhold context. Too little context fails faster and uglier than too much.

Technique · Open standard

Agent Skills

Folders of instructions that turn a general agent into a specialist — without bloating the context.

IS: A folder containing a SKILL.md file that teaches an agent how to perform a specific task — loaded only when that task comes up.
Use when: When a real, repeatable procedure exists, with rules of its own, that you currently explain by hand every time you need it.
Avoid when: When the instruction fits in two sentences, when the knowledge changes daily, or when what you actually need is access to a system — that is MCP, not a skill.

Concept

RAG vs enterprise RAG

Why the prototype that answers beautifully in one afternoon does not survive first contact with a real company.

IS: RAG is retrieving relevant passages and injecting them into the prompt. Enterprise RAG is the same idea plus permissions, freshness, conflict resolution, citable evidence and continuous evaluation.
Use when: Use RAG when the answer depends on knowledge that changes and the model does not have. Use the enterprise version when a wrong answer has consequences.
Avoid when: Do not use RAG when the knowledge is stable and small enough to fit in the prompt, nor when the real problem is that the information does not exist anywhere.