effective-writing
/effective-writing
Sprawdza akapit według kryteriów pisania dla czytelnika EW-1 do EW-10: rola, stawka, wyzwanie przekonań i podmioty i zakończenia zdań.
- Autor
- FASTMANMO ↗
- Repozytorium
- researchward ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 12.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 62
- Skrypty
- 0
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 12.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
Co robi i kiedy
Sprawdza akapit według kryteriów pisania dla czytelnika EW-1 do EW-10: rola, stawka, wyzwanie przekonań i podmioty i zakończenia zdań.
Oryginalny opis (EN)
Checks one paragraph against Larry McEnerney's writing-for-readers criteria (EW-1 to EW-10): does it give readers a role, name and challenge what they believe, show what is at stake for them, describe the contribution in words they can use, and put what they care about in sentence subjects and endings, with nothing that makes them slow down or reread. Fills a reader card first, shows the paragraph sentence by sentence, then marks each criterion met, partly met or not met with what the paragraph does, one fix for each gap and one priority fix. Use it whenever someone asks to check, improve, tig
Jak działa
- Scope
- 1. Choose the mode
- 2. Fill the reader card (G1)
- 3. Number the sentences
- 4. Run the criteria in order
- 5. Ground rules
- 6. Report (fixed format)
Podgląd SKILL.md
---
name: effective-writing
description: "Checks one paragraph against Larry McEnerney's writing-for-readers criteria (EW-1 to EW-10): does it give readers a role, name and challenge what they believe, show what is at stake for them, describe the contribution in words they can use, and put what they care about in sentence subjects and endings, with nothing that makes them slow down or reread. Fills a reader card first, shows the paragraph sentence by sentence, then marks each criterion met, partly met or not met with what the paragraph does, one fix for each gap and one priority fix. Use it whenever someone asks to check, improve, tighten or diagnose a paragraph, an abstract, an opening paragraph or any short piece of their own writing for its readers, or asks why a paragraph isn't landing. Load it afresh for every paragraph, even later in a conversation. For someone else's paragraph it gives the same table without fixes. For a whole paper's problem, stance and stakes use mcenerney-lens."
license: MIT
metadata:
version: "0.6.3"
author: "Mohammed Meligy, Mechanical and Aerospace Engineering, Arizona State University"
source: "Handout 2, Writing for Readers: Paragraph Analysis Criteria (docs/handouts in the repo)"
---
# Effective writing
Readers read in time, don't trust the writer, and stop as soon as a text stops being worth their time. Rules such as "use short sentences" or "avoid the passive" are techniques that only make sense once you know who reads the text and what it is for. This skill checks one paragraph against those readers. It follows Handout 2, *Writing for Readers: Paragraph Analysis Criteria*, compiled from McEnerney's lecture *Writing Beyond the Academy*. Criterion IDs (EW-1–EW-10), principles (P1–P7) and ground rules (G1–G6) are the handout's and never change.
## Scope
- **Covers:** one paragraph at a time, at sentence level (EW-1–EW-10).
- **Leaves to mcenerney-lens:** the paper's readers, problem, stance, stakes and contribution. If a McEnerney-lens report for this draft is already in the conversation, mark EW-1–EW-4 for its opening paragraph "covered by lens" and run EW-5–EW-10 only there. Don't repeat the lens's findings.
- **Leaves out:** a venue's rules, voice and layout.
- **Shared with mcenerney-lens:** EW-1 ↔ L12, EW-2 ↔ L4 and L5, EW-3 ↔ L6, EW-4 ↔ L8.
- **Several paragraphs:** run EW-1–EW-4 on each and EW-1–EW-10 on the one the user picks (default: the first), and say which.
## 1. Choose the mode
- **SELF-CHECK**: the user's own paragraph. Statuses, one fix move per criterion, one priority fix.
- **DISSECT**: someone else's paragraph (G6). The same table, with whether each criterion is met and what the paragraph does; no fix moves and no priority fix.
If it is unclear whose text it is, ask once; otherwise choose, and the report's header shows the choice.
## 2. Fill the reader card (G1)
| Field | Question |
|---|---|
| Readers | Who, specifically, will read this? A journal's community, a reviewer, a committee, a client |
| Current belief | What do they already think, assume or do that this paragraph touches? |
| What they value | What would make reading this worth their time? |
| Function | After this paragraph, what should they think, do or decide differently? |
| Reading conditions | At a desk or skimming? Will they reach the end of a long sentence? |
Fill it from what is already known: the user's message or a McEnerney-lens report's L1. Ask only for what is still missing, readers and function first. If no one can answer, state the assumption you used.
## 3. Number the sentences
S1, S2, … For several paragraphs, label them too: P1 S1, P2 S3. The report shows the paragraph with these numbers.
## 4. Run the criteria in order
Status key: STRONG does it · PARTIAL partly · MISSING not at all · N/A does not apply. The report prints them as Met, Partly met, Not met and Does not apply. The status lines apply each criterion; they are not printed in the handout.
**A. Value: does the paragraph give readers a reason to read?**
**EW-1 The reader's role in the opening**
- Ask: Does S1 give the reader a role, or only announce a topic?
- Check: Place S1 on the ladder: L0 says what the text is about (no role for the reader) · L1 says what it argues · L2 says what question it answers · L3 names the question the readers have, or ought to have.
- Red flags: "This paper/section/paragraph discusses, outlines, presents, describes…"; opening on background or definitions; the writer as the only actor.
- Fix: Start S1 from the readers' situation: what they think, need or asked.
- Status: STRONG at L3, or S1 starts from the readers' situation · PARTIAL at L1–L2 · MISSING at L0.
**EW-2 The readers' belief, named and challenged**
- Ask: Does the paragraph say what the readers currently think, and why that is incomplete or wrong? (Outside the academy: say what the readers need, and that this text supplies it.)
- Check: Find the sentence that states the community's assumption or practice, then the turn: however, but, unrealistic, it is claimed, incomplete, inconsistent.
- Red flags: The paragraph describes the world, not the readers' understanding of it. A turn word with nothing before it to turn against.
- Fix: Add one sentence that states the readers' view fairly, then make the turn.
- Status: STRONG the belief or practice is stated, then turned · PARTIAL a turn against a belief that is only implied · MISSING no turn against what the readers think, even if their practice is mentioned.
**EW-3 Stakes and urgency**
- Ask: Can these readers tell what it costs *them* if they ignore this, or what they gain if they act on it?
- Check: Name the cost in one phrase, as a cost to the readers on the reader card. Then circle the words that carry tension: fails, cannot, error, risk, crisis.
- Red flags: Neutral placeholders: situation, aspects, implications, issues, has been studied. A paragraph in which nothing is at risk.
- Fix: Replace the neutral noun with the concrete cost: what breaks, for whom, and how badly.
- Status: STRONG the paragraph says what the problem costs these readers in their own work (wrong predictions, wasted runs, a bad decision) · PARTIAL the cost is stated only as what the model or the physics does, even when it is "their" model, or only the benefit is stated · MISSING nothing is at risk. mcenerney-lens judges L6 the same way.
**EW-4 Words that signal value**
- Ask: Do the words that describe the contribution tell readers they can use it?
- Check: Circle every noun that names what is offered. Sort them into use words (tool, method, test, criterion, diagnostic, a model that predicts…) and display words (ideas, concepts, thoughts, discussion, overview, insights).
- Red flags: The contribution described only with display words, or only as an activity ("X is analyzed").
- Fix: Name what the reader can do with it.
- Status: STRONG named with use words · PARTIAL use and display words mixed · MISSING display words only, or an activity only · N/A the paragraph offers no contribution.
**B. Sentences: is the value where readers look for it?**
**EW-5 Focus: subjects carry what readers care about**
- Check: List the grammatical subject of each sentence, in order (in English the subject is the focus about four times in five). Do these readers care about each one? Do they stay on one line of interest?
- Red flags: Subjects that are the writer's apparatus (this study, the authors, the code, Section 3) when readers care about the phenomenon; subjects that jump between unrelated things.
- Fix: Move what the readers care about into subject position. Active or passive is chosen by focus, not by rule.
- Status: STRONG the subjects are what these readers care about, on one line of interest · PARTIAL some subjects are the writer's apparatus or jump between topics · MISSING no subject is something these readers care about.
**EW-6 Stress: endings carry what matters**
- Check: Read only the last few words of each sentence, in order. Is that the paragraph's message?
- Red flags: Endings that hold hedges, citations, locations ("in this context", "as shown in Fig. 3") or old information. A run of short sentences that stresses everything equally.
- Fix: Move new, important information to the end; let longer sentences build toward the main stress.
- Status: STRONG the endings, read in order, carry the paragraph's message · PARTIAL some endings hold method, old information or a hedge where new information should be · MISSING the endings are hedges, citations or locations, with no content in the stress position. MISSING only when every ending is a hedge, a citation or a location; if any ending holds content, even old information or method, it is PARTIAL.
**EW-7 Value along the way**
- Check: For each long sentence (about 25 words or more), mark the first point where the reader gains something. If the payoff comes only at the end, ask whether these readers will get there (reading conditions).
- Red flags: Several lines of setup before the point, in a venue read quickly or on screen.
- Fix: Bring an early payoff forward, or split the sentence. Shortness is not the goal.
- Status: STRONG every long sentence pays off along the way · PARTIAL some pay off only at the end · MISSING long sentences pay off only at the end, for readers who won't get there · N/A no long sentences.
**EW-8 Concision as processing time**
- Check: Mark sentences where the reader must re-sort who did what, hold the subject across a long interruption before the verb, or flip the focus to the thing they care about.
- Red flags: Active or passive chosen by rule; nominalizations that hide the actor; a short sentence that still forces a reread.
- Fix: Rebuild the sentence around the reader's focus, even if it gets a few words longer.
- Status: STRONG no sentence shows any of the Check's marks · PARTIAL one sentence does (who-did-what to re-sort, a subject held across an interruption of about 10 words or more before its verb, or a flipped focus), even if a careful reader gets through it · MISSING several do. Count the words between each subject and its verb; 10 or more is PARTIAL, however smooth the sentence feels.
**C. Reading process: does the paragraph keep readers reading?**
**EW-9 Reading friction**
- Check: Mark the first point where a target reader would slow down, and name its cause: a term these readers don't use, jargon that adds nothing for them, an ambiguity, or a missing link between sentences.
- Red flags: "It's the reader's job to read it." Jargon judged by a rule rather than by the readers.
- Fix: Fix the first friction point first. Keep the jargon these readers use and value; explain or cut the rest.
- Status: STRONG no friction point for these readers · PARTIAL one friction point · MISSING friction early enough to stop them. Count a friction point only where these readers would actually stop or reread; a link or pronoun an expert reader resolves at once is not one. A friction-free paragraph can still have no value.
**EW-10 Changing versus demonstrating**
- Check: Tag each sentence C (moves the reader's thinking or decision) or D (demonstrates knowledge, method or effort).
- Red flags: A paragraph of D sentences; method recital or background with no consequence for the reader; "It is well known that…" with nothing done to it.
- Fix: Connect each D sentence to a change (why should these readers care that you know this?), or cut it.
- Status: STRONG every sentence is C, or each D sentence is tied to a change · PARTIAL a mix of C and untied D · MISSING all D.
**Priority fix (self-check):** the one criterion that most limits value, usually the first of EW-1–EW-3 that isn't STRONG. Value comes before polish.
## 5. Ground rules
- **G1** No finding without the reader card. If the readers or the function are unknown, ask, or state the assumption you used.
- **G2** Never report a text rule as a finding on its own. Translate it into a reader effect: not "uses the passive voice" but "the subject of S3 is the apparatus, while these readers care about what it measured".
- **G3** Every finding cites sentence numbers.
- **G4** Judge concision by the reader's processing time, not by word count.
- **G5** Diagnose first; one fix move per criterion. Keep the writer's content, terminology, numbers and technical claims intact. Rewrite the paragraph only if the user asks.
- **G6** Someone else's paragraph: say whether each criterion is met and what the paragraph does, with sentence numbers, but give no fix move and no priority fix: it isn't the user's to revise.
## 6. Report (fixed format)
The report has one format, the same for every paragraph in a conversation, whatever was discussed in between. A reader who has never seen the handout must be able to follow it.
- **Every report, every time:** re-read this section first, even if this skill ran earlier in the conversation (invoke it again if its text is not in view). Build the report from this template and the paragraph, never from an earlier report. Requests made in discussion applied to that reply only. Anything asked besides the report goes after it, below a horizontal rule. If the user asks to change the report itself, change that one report and say so under the header; the next one returns to this template.
- **Labels:** criteria by number and plain-words name, exactly: 1- Opening gives readers a role · 2- Names and challenges what readers believe · 3- Shows what is at stake for readers · 4- Offers something readers can use · 5- Sentence subjects are what readers care about · 6- Sentence endings carry the key point · 7- Long sentences pay off early · 8- No sentence needs a reread · 9- Nothing slows readers down · 10- Changes what readers think, not only shows knowledge.
- **Plain words only:** no abbreviations and nothing that needs the handout: no IDs (EW-1, P3, G1), no ladder codes (L0–L3), no "C" or "D" tags, no STRONG, PARTIAL, MISSING or N/A (write Met, Partly met, Not met, Does not apply). Say what you mean instead: "S1 only names a topic", "S2 and S3 show what the writer knows but don't change what readers think". Sentence labels (S1, S2…) are fine, since the report shows them.
- **Hard limits:** ordinary chat formatting, never a code block, nothing before the header. The paragraph exactly as given, one sentence per line, with a blank line between sentences so each shows on its own line. The table has exactly three columns (Criterion · Met? · What the paragraph does) and all ten rows, 1- to 10-, in order. Each "What the paragraph does" cell at most 30 words, citing sentence labels. Fixes go below the table, one line per criterion that is partly met or not met, each at most 15 words. Before sending, check the columns, the ten rows, the labels, the word counts, a sentence label in every cell of the third column (even "Nothing is at risk in S1–S4"), exactly one action in each fix, and that no abbreviation slipped in.
**Effective-writing report** · Self-check *(or Dissect)*
**Reader card** (who the paragraph is for): readers · what they believe now · what they value · what they should think or do after reading · how they read (say which you assumed)
**The paragraph**
**S1** First sentence, exactly as written.
**S2** Second sentence, exactly as written.
…
| Criterion | Met? | What the paragraph does |
|---|---|---|
| 1- Opening gives readers a role | Not met | S1 only names a topic: "This section discusses…", so readers get no question of their own |
| 2- Names and challenges what readers believe | … | … |
| 3- Shows what is at stake for readers | … | … |
| 4- Offers something readers can use | … | … |
| 5- Sentence subjects are what readers care about | Partly met | Subjects: S1 "this section", S2 "subgrid-scale models"… The writer's frame, not the readers' problem |
| 6- Sentence endings carry the key point | … | Endings: S1 "…", S2 "…" |
| 7- Long sentences pay off early | Does not apply | S1–S4 are all short |
| 8- No sentence needs a reread | … | … |
| 9- Nothing slows readers down | … | … |
| 10- Changes what readers think, not only shows knowledge | Not met | S1–S4 show what the writer knows; none changes what readers think or do |
**How to fix**
- 1- Opening gives readers a role: start S1 from what the readers think or need.
- … one line for each criterion that is partly met or not met, in order.
**Priority fix:** the one criterion that most limits value, by number and name, and why.
**Untouched:** numbers, results and technical claims are not changed.
- **DISSECT:** the same three columns and all ten rows; leave out the How to fix list and the Priority fix line.
- **After a McEnerney-lens report on the same draft,** for its opening paragraph: keep all ten rows, but rows 1- to 4- read "Covered by the lens report" in the Met? column and the third column only names the lens field that covers it (as in "See the lens report: 12- Opening states the problem"), with no finding of its own; they get no fix.
- **Several paragraphs:** show each one, its sentences labeled P1 S1, P1 S2, …, and say which paragraph got all ten criteria.
Pliki w paczce (62)
- SKILL.md 17 KB
- evals/dissect-published-paragraph/case.yaml 1 KB
- evals/dissect-published-paragraph/prompt.md 1 KB
- evals/dissect-published-paragraph/graders/01-mode-is-dissect.md 1 KB
- evals/dissect-published-paragraph/graders/02a-three-columns.md 1 KB
- evals/dissect-published-paragraph/graders/02b-no-fixes.md 1 KB
- evals/dissect-published-paragraph/graders/03-all-criteria-located.md 1 KB
- evals/dissect-published-paragraph/graders/04-traces-the-moves.md 1 KB
- evals/dissect-published-paragraph/graders/05a-paragraph-shown.md 1 KB
- evals/dissect-published-paragraph/graders/05b-subjects-and-endings.md 1 KB
- evals/dissect-published-paragraph/graders/06-chat-format.md 1 KB
- evals/dissect-published-paragraph/graders/07-plain-words.md 1 KB
- evals/self-check-version-a/case.yaml 1 KB
- evals/self-check-version-a/prompt.md 1 KB
- evals/self-check-version-a/graders/01-reader-card-from-prompt.md 1 KB
- evals/self-check-version-a/graders/08-no-rewrite.md 1 KB
- evals/self-check-version-a/graders/20-paragraph-shown.md 1 KB
- evals/self-check-version-a/graders/21-three-columns.md 1 KB
- evals/self-check-version-a/graders/22-ten-rows-in-order.md 1 KB
- evals/self-check-version-a/graders/23-how-to-fix.md 1 KB
- evals/self-check-version-a/graders/24-plain-words.md 1 KB
- evals/self-check-version-a/graders/25-chat-format.md 1 KB
- evals/self-check-version-a/graders/26-every-row-cites-a-sentence.md 1 KB
- evals/self-check-version-a/graders/27-one-fix-per-criterion.md 1 KB
- evals/self-check-version-a/graders/30-status-1.md 1 KB
- evals/self-check-version-a/graders/31-status-2.md 1 KB
- evals/self-check-version-a/graders/32-status-3.md 1 KB
- evals/self-check-version-a/graders/33-status-4.md 1 KB
- evals/self-check-version-a/graders/34-status-5.md 1 KB
- evals/self-check-version-a/graders/35-status-6.md 1 KB
- evals/self-check-version-a/graders/36-status-7.md 1 KB
- evals/self-check-version-a/graders/37-status-8.md 1 KB
- evals/self-check-version-a/graders/38-status-9.md 1 KB
- evals/self-check-version-a/graders/39-status-10.md 1 KB
- evals/self-check-version-a/graders/40-priority.md 1 KB
- evals/self-check-version-a/graders/41-opening-reason.md 1 KB
- evals/self-check-version-a/graders/42a-subjects-name-this-section.md 1 KB
- evals/self-check-version-a/graders/42b-endings-name-dns-data.md 1 KB
- evals/self-check-version-b/case.yaml 1 KB
- evals/self-check-version-b/prompt.md 1 KB
- evals/self-check-version-b/graders/06-no-rewrite.md 1 KB
- evals/self-check-version-b/graders/20-paragraph-shown.md 1 KB
- evals/self-check-version-b/graders/21-three-columns.md 1 KB
- evals/self-check-version-b/graders/22-ten-rows-in-order.md 1 KB
- evals/self-check-version-b/graders/23-how-to-fix.md 1 KB
- evals/self-check-version-b/graders/24-plain-words.md 1 KB
- evals/self-check-version-b/graders/25-chat-format.md 1 KB
- evals/self-check-version-b/graders/26-every-row-cites-a-sentence.md 1 KB
- evals/self-check-version-b/graders/27-one-fix-per-criterion.md 1 KB
- evals/self-check-version-b/graders/30-status-1.md 1 KB
- evals/self-check-version-b/graders/31-status-2.md 1 KB
- evals/self-check-version-b/graders/32-status-3.md 1 KB
- evals/self-check-version-b/graders/33-status-4.md 1 KB
- evals/self-check-version-b/graders/34-status-5.md 1 KB
- evals/self-check-version-b/graders/35-status-6.md 1 KB
- evals/self-check-version-b/graders/36-status-7.md 1 KB
- evals/self-check-version-b/graders/37-status-8.md 1 KB
- evals/self-check-version-b/graders/38-status-9.md 1 KB
- evals/self-check-version-b/graders/39-status-10.md 1 KB
- evals/self-check-version-b/graders/40-priority.md 1 KB
- evals/self-check-version-b/graders/41-stakes-reason.md 1 KB
- evals/self-check-version-b/graders/42-focus-reason.md 1 KB
Wersje
W bazie nie ma innych skilli o nazwie effective-writing.
Migawka bazy z 13.11.2026 · wersja paczki 0bdcea2b65bc (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 100mcenerney-lens
Czyta pracę naukową lub szkic przez soczewkę wartości dla czytelnika, wskazując odbiorców, obietnicę, stawkę i dokładne miejsca w tekście. - 90writing-shape
Kształtuje dokument źródłowy akapit po akapicie, bez modyfikowania go. - 100doc-subtract
Streszcza dokument, zamieniając każdy akapit w jedno zdanie, a potem wyciągając tezę, kluczowe punkty i wniosek.