Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Use when drafting or formatting thesis/dissertation sections: apply the exact word-count, numbering, and reference conventions that academic submission guidelines require.
.claude/skills/thesis-dissertation-compliance/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 1 |
| Model | Lift | Δ tokens | Δ turns | Cases | Verified |
|---|---|---|---|---|---|
| gemini-3.5-flashbest | +50% | — | 0% | 24 | 86d ago |
| gemini-3.6-flash | +42% | +144% | 0% | 24 | 54d ago |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | — | — |
| case-17 | ✗→✓ | ▲ Improved | — | — |
| case-18 | ✗→✓ | ▲ Improved | — | — |
| case-02 | ✗→✓ | ▲ Improved | — | — |
| case-14 | ✗→✓ | ▲ Improved | — | — |
Enforces the structural, word-count, numbering, citation, and unit conventions of a formal thesis/dissertation submission guideline. Apply whenever drafting or formatting any thesis or dissertation section (abstract, acknowledgements, keywords, body chapters, tables/figures, references) — these are arbitrary house rules a general writing default does not follow.
objectives, methods, results, conclusions.
jokey. Name the specific people (supervisor, colleagues), the department, and any funding body.
research topic. Placed immediately after the abstract.
write "polymerase chain reaction", never "PCR"; "magnetic resonance imaging", never "MRI".
parentheses — then use the abbreviation thereafter. Standard SI units are exempt.
Chapter.Sequence in Arabic numerals. The first table in Chapter 3 isTable 3.1, the second Table 3.2; the first figure in Chapter 5 is Figure 5.1.
in Table 3.1").
heading abbreviations are explained in a footnote. Figures get a legend/caption.
al. 2019").
on each page.
(Surname Year) —never numbered [1] and never superscript.
for the same compound.
Abstract length & content. BEFORE: a tidy 180-word paragraph that just summarizes the topic. AFTER: a 300–500-word abstract whose sentences walk through objectives → methods → results → conclusions in that order.
Abbreviation in the abstract. BEFORE: "We used RT-PCR to detect HBV in serum." AFTER: "We used reverse transcription polymerase chain reaction to detect hepatitis B virus in serum." (No acronym anywhere in the abstract.)
Keyword count. BEFORE: a comma list of 8 keywords, some full phrases. AFTER: exactly 4 — "coral reef bleaching; ocean acidification; thermal stress; calcification" — placed right after the abstract.
Acknowledgements length/tone. BEFORE: "Thanks to everyone who helped, you know who you are." AFTER: a 300–500-word passage naming Dr. Patel (supervisor), the lab colleagues, the National Research Council (funding), and family, in a sincere, rigorous tone.
Table numbering. BEFORE: "Table 1 shows enzyme activity… Table 2 shows protein yield." (Chapter 3). AFTER: "As shown in Table 3.1, enzyme activity… Table 3.2 reports protein yield." Each referenced before it appears; each has a brief title.
Figure numbering. BEFORE: "Figure 1: survival rates… Figure 2: tumor growth." (Chapter 5). AFTER: "Figure 5.1 plots survival rates by treatment group… Figure 5.2 shows tumor growth over time." Each has a legend; each is cited in the text first.
Adapted figure source. BEFORE: a reproduced metabolic-pathway figure with no attribution. AFTER: the figure numbered Figure 4.1, with "Source: adapted from Garcia and Lee (2021)" beneath it and a legend.
In-text citation style. BEFORE: "Soil microbial diversity declines under monoculture 12]." AFTER: "Soil microbial diversity declines under prolonged monoculture (Okafor 2020)."
Reference list. BEFORE: 6 sources in numbered order of first mention. AFTER: ≥10 sources ordered alphabetically by first author's surname, each in Harvard author-date form.
Units. BEFORE: "glucose ranged from 90 mg/dL to 5.0 g/L." AFTER: "glucose ranged from 5.0 to 7.8 mmol/L" — one consistent unit, capital L.
other abbreviations do.
your own original data does not. When in doubt about "adapted," cite.
Table 4.1, not "Table 4" or "Table 1".
Table 3.1 and Figure 3.1 — figures and tables are numbered separately.
use; the prohibition is specifically on the title and abstract.
Don't stop at a 150–250-word teaser.
HBV) slip into them.
Chapter.Sequence. Don't number globally (Table 1, Table 2).textual call-out.
source.
(Surname Year). Don't use [1] or superscript numbers.citation appearance or list fewer than 10.
lowercase "l".
[n] or Vancouver-style citations instead of Harvard author-date.(Surname Year); reference list alphabetical; ≥10 entries.| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-17 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-09 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-14 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-07 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-23 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-22 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-06 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-03 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-08 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-24 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
DecimalAI ran this skill against gemini-3.5-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 24 cases were attempted. The headline lift of +42 percentage points is the difference between those two pass rates over the 24 comparable cases.
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 | 6/27/2026 | +50% |
Other measured skills in the registry, with their headline benchmark lift.