The SCQA Communication Coach skill for Claude

The problem this skill solves is spelled out in its own activation rules. It kicks in for requests like structure my message and build a pitch, but also for frustrations: I get straight to the point and nobody understands why, my email never gets a reply, my presentation starts too slowly, they don't get the problem, I can't get sign-off on my proposal. Say something like that to Claude while the skill is loaded and it activates automatically. Those aren't made-up complaints, they're the moments where a message without structure falls apart.

It's written for anyone who has to convey something that a decision or a change in behaviour needs to follow: an email with bad news, a briefing note for management, an investor pitch, a policy document, the announcement of a change. It always delivers the written-out text rather than advice to try it yourself, and it never invents a reason, a figure or an event to make the complication more dramatic. It belongs to the AI and automation service and, like the rest of the skill library, can be downloaded without an account and without a sales email.

There's one thing you should know beforehand, and the file states this principle clearly: SCQA is not a template to lay over a text, it's a test. If you can't fill in the four parts, the message itself isn't finished yet, and writing won't help. And it will simply tell you that.

The opening is about the sender. According to the file, the situation is a factual description of how things stand now, one the reader agrees with immediately. No introduction about yourself, no list of context: the shared starting point. If the reader thinks that's not right at the very first sentence, you've lost them before you've begun. That's why the skill tests every situation for agreement before it goes any further.

There's no complication. The complication is what changes, clashes or goes wrong, and the file calls it the engine of the whole piece. Without a complication there's no reason to keep reading, because then nothing is actually going on. Many texts that don't land have this flaw: they repeat the situation in other words and call that urgency.

The question never gets formulated. The question is what the complication logically raises in the reader's mind. It usually doesn't appear literally in the text, but it does steer everything else: any paragraph that doesn't serve the answer to that question can go. That's why the skill always writes the question out in full, even if it doesn't make it into the final text, and uses it as a filter.

The answer arrives too late or stays vague. The answer is your conclusion or recommendation in one sentence, and immediately the top of your pyramid: everything after that is supporting evidence. An answer without a verb and without a date isn't an answer, it's a direction, and nobody can say yes to that.

The proportions are off. One writing rule from the file that pricks straight through a lot of texts: the situation is never longer than the complication. If you need three paragraphs of context, you've picked the wrong situation.

In diff form, with the plus lines taken literally from the example in the skill: what comes out and what replaces it.

Then comes the theory: the four parts with their definitions, the link to the Pyramid Principle with the MECE rule, the four order variants that Minto herself describes, and the shortened SCR form for short messages.

Then the method: seven steps the skill always takes. It first establishes who the reader is, what they already know and what they need to do after reading. It fills in the four parts separately and shows them on their own, before any running text appears. It tests the situation for agreement and the complication for authenticity. It spells out the question literally and uses it as a filter for everything that follows. It builds two to four arguments under the answer following the pyramid and tests them for MECE. And it writes the whole thing out in the requested format, with an alternative order included if the situation calls for it.

Before anything is written, the skill runs through a mandatory seven-point checklist: your message or raw notes, the reader and what they already know, what you want from the reader, what has changed or gone wrong, the format and the length, the relationship between you and the reader, and what argues against your proposal. It only asks about what's missing, in a single round, and then continues with a marked assumption. So you don't get a form, at most a few targeted questions.

The output always has the same seven blocks, in the same order: reader and goal, SCQA at a glance, the test on the four parts, the pyramid under the answer, the written-out piece, an alternative order, and a closing block with what isn't backed up and which objection is still open. That last block is typical of the file: the skill names the weak spots in its own piece before your reader does. In the example from FIG.01, that's the missing argument for two days instead of one, and the unanswered objection that pitches have also been won online.

A striking number of rules are about language: clear, direct British English, no dashes in running text, short sentences alternated with long ones, concrete over abstract, figures over adjectives, and a ban list of hollow phrasing. There is also a separate six-step review mode for existing work, more on that in the FAQ. At the bottom are three source references, all traceable to Barbara Minto and the consulting practice at McKinsey. If you want to learn to write an instruction file like this yourself: the structure is exactly what we explain in writing SKILL.md files NL.

The introduction and the pyramid belong together. SCQA takes the reader from a shared starting point to your conclusion, and that conclusion is immediately the top of the pyramid: below it hang two to four arguments of the same kind, and below each argument the facts. The arguments have to be MECE: they don't overlap and together they cover the question. This skill builds both, but if you mainly want to train the supporting-evidence side, there's the separate Pyramid Principle Writer skill.

The structure is a test, not a trick. That's the most practical lesson from the file. If you can't formulate the situation as something the reader says yes to, you don't know your reader well enough. If you can't point to a complication, you have no message. If you can't get the answer into one sentence, you haven't made a decision yet. The structure exposes that before you start writing, and that saves you exactly the hours you'd otherwise lose shuffling paragraphs around.

Where SCQA stops, other structures begin. The file is open about this: SCQA is far from a fit everywhere. For ad copy where the hook matters more than the lead-in there's the AIDA Copywriter skill, for a brand story where the customer plays the lead role the StoryBrand Writer skill, and if you want a message to stick rather than just be understood, look at the Made to Stick skill. Each of those structures solves a different problem; SCQA is the structure for the message that a decision has to follow.

Even without installing the skill, you can use these principles to improve your next email: start with a sentence your reader says yes to, name what's changing, and put your answer in one sentence with a verb and a date.

After that, using it is simple: paste your message or your raw notes and say what it needs to become. In a single round it asks what's still missing, and then it gets to work. If you get stuck anywhere, the full explanation per environment, including screenshots and common mistakes, is in installing Claude skills NL. And if you'd rather read the wider story about working with AI first, start in the knowledge base.

There are also texts you're better off leaving it for, and the file names them itself. SEO blog articles are built on search intent and follow the searcher's question, not the logic of an argument; how that does work you can read at improving your SEO. Ad copy lives on the hook in the first second, and SCQA's lead-in is too patient for that. And narrative texts want to build tension rather than lay it on the table straight away. For those cases other structures exist, and some of them are available as a separate skill in the library.

The most honest limit is already in this page's hero, but it deserves repeating: the skill makes your message understandable, not automatically welcome. A reorganisation doesn't get more fun from a good structure, and a weak proposal doesn't get strong from a tight pyramid. What the structure does do: make sure the reader understands exactly what you're saying and what you're asking, so the conversation afterwards is about the content and not the confusion. If you're not sure beforehand whether your proposal survives the test itself, run the Pre-Mortem Analyst skill over it first: it looks for the ways your plan could fail before you announce it.

A good structure only helps if your message reaches the right people. For an email or a briefing note, you choose those people yourself, and then the skill is enough. For a page on your site, the search engine has a say too, and then what counts first is whether you show up at all. The free SEO scan shows where you stand in a few seconds, and the SEO improvement service takes it from there if there's more to do than you want to handle yourself. In the guide writing models compared you can see how this model relates to the others.