angushushu/shu-academic-writing-skill / core/engineer-argument
engineer-argument
VerifiedDiagnose why a draft will not land with its readers, using Larry McEnerney's framework — value over clarity, reader-driven prose, instability and cost, community codes. Use when someone shares a paper, abstract, introduction, dissertation chapter, literature review, grant or research proposal, or conference submission and wants feedback, review, critique, or "does this work?" — and when they ask why their writing gets rejected, desk-rejected, ignored, or called boring or unconvincing, how to make an argument compelling, whether their contribution is significant enough, or how to write a stronger intro or abstract. This is the top-down structural review: value, problem, movement, code, and the sentence patterns that follow. For a pure line-level pass on prose whose argument is already settled, read `core/polish-prose/SKILL.md` instead. Default behavior is diagnosis and coaching, never rewriting.
| File | Kind | Size |
|---|---|---|
| SKILL.md | skill md | 12.8 KB |
| references/argument-patterns.md | markdown | 4.2 KB |
| references/diagnostics.md | markdown | 10.6 KB |
| references/drills.md | markdown | 6.1 KB |
| references/genres.md | markdown | 9.2 KB |
| references/principles.md | markdown | 16.2 KB |
| references/sentence-patterns.md | markdown | 5.1 KB |
references/drills.md
View on GitHub# Drills
A review fixes one draft. Drills change a writer. Assign **two or three**, matched to the
habits you diagnosed — a list of a dozen exercises gets skipped entirely.
Each drill states the habit it treats, what the author does, and how they know it worked. Give
the author the drill in their own terms; don't just paste the name.
These are the structural drills — value, problem, movement, code. For sentence-level drills
(subjects and verbs, stress positions, hedge stacking), the `core/polish-prose` module carries
the full set; two of the most useful are reproduced at the bottom here so a review can assign
one without a handoff.
---
## 1. The value sentence
**Treats:** topic-writing, no stated change, "this paper examines."
Before writing or revising anything, complete on paper:
> Readers of ______ believe ______. After this, they will believe ______, which matters
> because ______.
Rules: the first blank names a venue or a nameable group, never "the general reader." The
third blank contains a claim, not a topic. If the fourth blank takes more than a minute, the
problem isn't writing — it's that the value hasn't been found yet, which is worth knowing
before spending a week on prose.
**Done when:** the four blanks fit in one breath and a colleague in the field nods rather than
asking "so?"
## 2. The hostile "so what?"
**Treats:** gap-framing, unmotivated significance.
Take your significance claim. Write "So what?" under it and answer in one sentence. Then write
"So what?" under *that* answer. Three rounds. The third answer is usually the real stakes, and
usually isn't in the draft at all.
**Done when:** an answer arrives that names a consequence for someone's actual work.
## 3. The instability count
**Treats:** topic tours, flat openings.
Highlight every contrastive and problem word in your introduction: *but, however, yet,
although, despite, whereas, surprisingly, anomalous, inconsistent, fails to, unresolved, at
odds.* Count them. Then count the sentences.
Interpret honestly: a methods section should score near zero. An introduction with one marker
in twenty sentences is not being read.
**Done when:** you can point to the sentence where a reader first learns that something they
believe is in trouble — and it's on page 1.
## 4. The topic-string read
**Treats:** accumulation, list-structure, circling.
Copy the first six words of every paragraph into a separate document and read that list alone.
It's the skeleton your skimming reader sees.
Ask: does it advance, or does it inventory? Would shuffling it lose anything? Does the same
noun appear at the head of five consecutive paragraphs?
**Done when:** the list reads as a sequence of moves rather than a table of contents.
## 5. The skim simulation
**Treats:** buried claims, late problems, unpaid promises.
Give someone in your field the title, abstract, and first sentence of each paragraph — nothing
else — and thirty seconds. Then ask them what your claim is and why it matters. Their answer,
not your intention, is what the paper currently communicates.
No colleague available? Set the draft aside for two days and do it yourself with a timer.
**Done when:** their summary matches your value sentence from drill 1.
## 6. Code harvest
**Treats:** under-coding, addressing the wrong community, outsider signalling.
Take three recent articles from your exact target venue. Circle every word that repeats across
all three and would look strange in a newspaper. That's the code.
Then check your draft: which of those terms appear? Where are you using ordinary language for
something your field names precisely? Where are you defining something all three articles
assume?
**Done when:** you can list ten code terms for your venue and account for each one's presence
or absence in your draft.
## 7. The deletion test
**Treats:** accumulation, comprehensiveness habit, "showing the work."
Pick three middle paragraphs. For each, ask: what breaks if this is gone? Not "what is lost" —
what *breaks*. If the answer is "nothing breaks, but it's good material," it's function-1
writing that survived into draft 2.
**Done when:** you've actually deleted something you liked.
## 8. Two drafts, deliberately
**Treats:** polishing a thinking-draft, structural paralysis.
Split the work explicitly. Draft 1: write fast, do not edit, stop when you know what you
think. Then close it. Draft 2: on a blank page, write the value sentence, the instability, the
cost, and the claim — then build the structure from those, pulling material from draft 1 as
raw supply rather than as a starting text.
The hard part is not reopening draft 1 first. Its architecture will capture you.
**Done when:** draft 2's paragraph order differs substantially from draft 1's.
## 9. Read your field's rejections
**Treats:** vague sense of what "works," reliance on general advice.
Find two papers in your area you found compelling and two you abandoned halfway. For each, mark
where you engaged or quit, and why. Four papers is enough to see the pattern.
This builds the thing no writing guide can transfer: a model of your specific readers, derived
from your own reading behavior.
**Done when:** you can state one structural move that reliably makes you keep reading in your
field.
---
## Two sentence-level drills, for when a review needs one
## 10. Subjects and verbs
**Treats:** nominalization clusters, hidden agents, abstract subjects.
Take one page. Underline the grammatical subject and main verb of every sentence. Read only
those pairs aloud.
You should hear characters doing things. If you hear "analysis was conducted," "consideration
is given," "the relationship exists" — the actors are missing, and the reader is reconstructing
them on your behalf.
**Done when:** most subject-verb pairs name someone or something doing something.
## 11. The claim in twenty words
**Treats:** overlong central claims, buried thesis, hedge stacking.
Write your paper's main claim in twenty words or fewer, as a single sentence, with no hedges.
Then decide which hedges to add back and name what each one is protecting against.
**Done when:** the twenty-word version is defensible and you can say precisely what each
restored qualifier does.
Related skills
| Skill | Stars | Updated | Works with | Status | Install |
|---|---|---|---|---|---|
| reverse-engineer-anythingmorluto/reaReverse engineer native, managed, Electron/JavaScript, packaged, firmware, and browser targets with REA, and build agent environments (AgentEnv bundles) of web applications from a live instance or a capture, and of Linux desktop apps with a task from a request. Use shipped-artifact or requested runtime evidence to explain features, compare versions, decompile code, or guide a reconstruction. Skip REA for ordinary source-repository architecture analysis. | Stars: 472k | Updated | Claude CodeCodex | Verified | Install |
| diagnosing-superpowersobra/superpowersUse when a superpowers session went wrong and your human partner wants to know why — repeated work, ignored plans, stumbles, poor results, a skill that didn't fire, "it took too long", "why is it so expensive", "what is it doing" — or wants to build a bug report for the superpowers maintainers, for the current session or a past one identified by id or path, on any harness. | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| dispatching-parallel-agentsobra/superpowersUse when facing 2+ independent tasks that can be worked on without shared state or sequential dependencies | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |
| executing-plansobra/superpowersUse when executing an implementation plan in the current session as the implementer yourself — your human partner chose inline execution, or no subagent tool is available | Stars: 322k | Updated | Claude CodeCodex | Verified | Install |