Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when formatting a dissertation/thesis section: number tables and figures per-chapter (Table 3.1), cite Harvard author-date, write liter as capital L, and keep the abstract under 500 words with no citations or abbreviations.
.claude/skills/dissertation-formatting-conventions/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 69 |
| Model | Lift | Δ tokens | Δ turns | Cases | Verified |
|---|---|---|---|---|---|
| gemini-3.6-flashbest | +18% | +140% | 0% | 22 | 54d ago |
| gemini-3.5-flash | pending re-run | — | |||
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-15 | ✗→✓ | ▲ Improved | — | — |
| case-08 | ✗→✓ | ▲ Improved | — | — |
| case-19 | ✗→✓ | ▲ Improved | — | — |
| case-20 | ✗→✓ | ▲ Improved | — | — |
| case-07 | ✓→✓ | = Same ✓ | — | — |
Enforces the formatting, structural, and stylistic conventions of a science dissertation/thesis. Apply whenever drafting, formatting, or reviewing any thesis section — front matter, chapters, references, tables/figures, abbreviations, or reported units. These rules override the model's common defaults (global figure numbering, numbered citations, lowercase liter, citations in the abstract).
Table 3.1, Table 3.2, Figure 3.1.The first number is the chapter; the second restarts at 1 in every new chapter.
Table 3.1, not Table III.A).Table 5.1 regardless of howmany tables preceded it in Chapters 1–4.
Table 4.1 and Figure 4.1.(as shown in Table 3.1). An unreferenced object is not allowed.
each continued page.
Source: Smith, 2019(verbatim copy) or Adapted from Smith, 2019 (modified).
(Johnson and Lee, 2018).et al.: (Garcia et al., 2021) — do not list every surname.(Patel, 2017).[1], ¹, (1)).appearance, and never numbered.
high-performance liquid chromatography (HPLC). Use the abbreviation alone thereafter.
of measurement (kg, mL, °C) are exempt and are not defined or listed.
37 °C.L: mL, L, mmol/L, g/L, U/L — never lowercase l(ml, mmol/l, u/l are wrong).
mmol/L, g/L); conventional clinical units (mg/dL, mg/mL) are acceptable.mg/dL and mmol/Lfor the same compound.
the lists of tables and figures each carry page numbers per entry.
a centered footer (department, faculty, submission date).
the supervisor(s).
rationale/gap and a list of general and specific objectives.
Numbering (per-chapter).
Table 1, Table 2, Figure 1.Table 3.1, Table 3.2, Figure 3.1.Numbering does not carry across chapters.
Table 9.Table 5.1 — numbering restarts at 1 in the new chapter.Caption placement.
Reproduced figure.
Figure 1. Pathway diagram.Figure 4.1. Pathway diagram. Source: Smith, 2019. placed below the figure, and referred toin the text (see Figure 4.1).
In-text citation form.
Three-or-more authors.
(Garcia, Smith, Wong, Ahmed and Rossi, 2021).(Garcia et al., 2021).Reference ordering.
Abstract length and content.
…consistent with the key prior study (Lee, 2017).citation removed.
Abbreviation in the abstract / title.
"high-performance liquid chromatography" with no parenthetical abbreviation.
Abbreviation in the body.
added to the List of Abbreviations.
Units — capital L.
Temperature.
Unit consistency.
mg/dL in one table and mmol/L in another.mmol/L) everywhere.Front-matter lists.
kg, mL, °C are used directly, neverdefined at first use, never placed in the List of Abbreviations. Only non-standard term abbreviations (HPLC, ELISA) are defined and listed.
5 mL) are allowed;the no-abbreviation rule targets term abbreviations, not unit symbols.
headings at the top of each continued page rather than splitting silently.
Adapted from … when you changed rows/values, Source: … for averbatim reproduction. Both still get a per-chapter number and a body reference.
Johnson and Lee, 2018); et al. startsonly at three authors.
in-text form stays author-date, never a number.
Table 3.1). Don't number globally (Table 1, Table 7).(Author, Year). Don't write [1] or a superscript number.or abbreviate inside it.
L. Don't write lowercase l (ml, mmol/l).mg/dL and mmol/L for the same compound.Table 1, Figure 2) instead of per-chapter.[1] (IEEE/Vancouver habit) instead of Harvard author-date.ml, mmol/l) — the convention requires capital L.…by ELISA).Table 3.1, Figure 3.1).Source: / Adapted from line.et al. for 3+; never [1].L; temperature in °C; one unit per analyte.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-03 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-06 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-09 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-08 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-14 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-17 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-01 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-07 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-22 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted. The headline lift of +18 percentage points is the difference between those two pass rates over the 22 comparable cases. 1 case got worse with the skill loaded, and it is included in that figure.
The per-case answers from this run were removed by the retention sweep, so the case table below shows the verdicts without the text either arm produced. The counts above were recorded at the time and are unaffected. Answers are now kept for 180 days.
| Model | Method | Date | Lift |
|---|---|---|---|
| gemini-3.5-flash | verified | 7/9/2026 | +42% |
Other measured skills in the registry, with their headline benchmark lift.