Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Guide to writing effective research paper introductions
.claude/skills/brycewang-stanford-introduction-writing-guide/SKILL.md| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-01 | ✗→✓ | ▲ Improved | 84% | 0% |
| case-11 | ✗→✓ | ▲ Improved | 362% | 0% |
| case-04 | ✓→✓ | = Same ✓ | 51% | 0% |
| case-07 | ✓→✓ | = Same ✓ | 144% | 0% |
| case-12 | ✓→✓ | = Same ✓ | 109% | 0% |
Write compelling research paper introductions using the CARS (Create A Research Space) model with structured approaches for establishing context, identifying gaps, and motivating your study.
John Swales' Create A Research Space (CARS) model is the most widely used framework for structuring academic introductions. It consists of three rhetorical "moves":
| Move | Purpose | Typical Length | |------|---------|---------------| | Move 1: Establishing a territory | Show the research area is important and active | 2-4 paragraphs | | Move 2: Establishing a niche | Identify a gap, question, or problem | 1-2 paragraphs | | Move 3: Occupying the niche | State what your paper does to address the gap | 1-2 paragraphs |
Signal that the topic is important, interesting, or well-established:
Pattern phrases:
- "X has attracted considerable attention in recent years..."
- "The role of X in Y is well established..."
- "X is a fundamental aspect of..."
- "Understanding X is critical for..."
- "Recent advances in X have transformed..."Summarize the current state of knowledge:
Pattern phrases:
- "Previous studies have demonstrated that..."
- "It is widely accepted that X leads to Y..."
- "Several approaches have been proposed to address X, including..."
- "Research in this area has traditionally focused on..."Cite specific studies to build the scholarly context:
Example paragraph:
"Smith et al. (2020) demonstrated that X improves Y by 30%.
Building on this finding, Jones (2021) extended the approach to Z domain,
reporting similar gains. Meanwhile, Chen and Lee (2022) proposed an
alternative framework that addresses the scalability limitations of
earlier methods."Tips for Move 1:
This is the most critical move. You must convince the reader that there is a problem worth solving.
Pattern phrases:
- "However, these approaches suffer from..."
- "Despite these advances, X remains poorly understood..."
- "A major limitation of existing methods is..."
- "These findings have been contradicted by..."Pattern phrases:
- "To date, no study has examined..."
- "Little attention has been paid to..."
- "The relationship between X and Y has not been explored..."
- "A comprehensive analysis of X is still lacking..."Pattern phrases:
- "An open question is whether X can be applied to..."
- "It remains unclear how X affects Y under conditions Z..."
- "This raises the question of..."Pattern phrases:
- "Following the approach of Smith (2020), we extend..."
- "Building on recent advances in X, this paper..."Tips for Move 2:
Pattern phrases:
- "In this paper, we propose..."
- "The present study aims to..."
- "This work introduces a novel approach to..."
- "We address this gap by..."Pattern phrases:
- "Our results demonstrate that..."
- "We show that X outperforms Y by Z%..."
- "The key finding is that..."Example:
"The remainder of this paper is organized as follows. Section 2
reviews related work. Section 3 describes our methodology. Section 4
presents experimental results. Section 5 discusses implications and
limitations. Section 6 concludes."markdown[Move 1A: Centrality claim - 1-2 sentences] [Topic] is a critical challenge in [field], with applications in [area 1], [area 2], and [area 3]. [Move 1B-1C: Background and review - 2-3 paragraphs] Previous work has established that [known fact]. Smith et al. (2020) showed [finding 1]. Jones (2021) extended this to [finding 2]. More recently, Chen (2022) demonstrated [finding 3]. [Move 2: Gap identification - 1 paragraph] However, existing approaches have several limitations. First, [limitation 1]. Second, [limitation 2]. To date, no work has addressed [specific gap]. [Move 3A: Purpose statement - 1-2 sentences] In this paper, we propose [method/framework] to address [gap]. Our approach differs from prior work in [key difference]. [Move 3B: Key findings - 1-2 sentences] Our experiments on [benchmark] demonstrate that [main result], achieving [quantitative improvement] over the state of the art. [Move 3C: Structure outline - optional, 1-2 sentences] The rest of this paper is organized as follows...
| Mistake | Problem | Fix | |---------|---------|-----| | Too broad opening | "Since the dawn of time..." | Start at the field level, not civilization level | | No clear gap | Reader does not understand why the paper is needed | State the gap explicitly in 1-2 sentences | | Overpromising | Claims too broad relative to actual contribution | Use hedging: "we investigate" not "we solve" | | Citing too few papers | Appears unaware of related work | Cite 15-25 papers in a typical introduction | | Too long | Buries the contribution | Aim for 1-2 pages (conference) or 2-4 pages (journal) | | Jargon overload | Inaccessible to non-specialist reviewers | Define key terms on first use |
| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-01 | fail→pass | 13,311 | 24,250 | +82% | 1 | 1 | 0% | 2,128 | 3,921 | +84% | 0 | 0 | — |
case-02 | fail→fail | 7,819 | 8,391 | +7% | 1 | 1 | 0% | 1,281 | 2,997 | +134% | 0 | 0 | — |
case-03 | fail→fail | 14,739 | 12,572 | -15% | 1 | 1 | 0% | 1,904 | 3,259 | +71% | 0 | 0 | — |
case-04 | pass→pass | 13,994 | 11,387 | -19% | 1 | 1 | 0% | 2,315 | 3,487 | +51% | 0 | 0 | — |
case-05 | fail→fail | 4,814 | 6,765 | +41% | 1 | 1 | 0% | 734 | 2,536 | +246% | 0 | 0 | — |
case-06 | fail→fail | 9,909 | 9,055 | -9% | 1 | 1 | 0% | 1,576 | 2,839 | +80% | 0 | 0 | — |
case-07 | pass→pass | 8,308 | 10,900 | +31% | 1 | 1 | 0% | 1,271 | 3,105 | +144% | 0 | 0 | — |
case-08 | fail→fail | 16,508 | 15,713 | -5% | 1 | 1 | 0% | 2,178 | 4,045 | +86% | 0 | 0 | — |
case-09 | fail→fail | 11,568 | 10,037 | -13% | 1 | 1 | 0% | 1,688 | 3,115 | +85% | 0 | 0 | — |
case-10 | fail→fail | 6,940 | 7,009 | +1% | 1 | 1 | 0% | 1,170 | 2,771 | +137% | 0 | 0 | — |
case-11 | fail→pass | 4,807 | 10,471 | +118% | 1 | 1 | 0% | 687 | 3,174 | +362% | 0 | 0 | — |
case-12 | pass→pass | 6,329 | 5,121 | -19% | 1 | 1 | 0% | 1,183 | 2,469 | +109% | 0 | 0 | — |
case-13 | fail→fail | 10,189 | 6,495 | -36% | 1 | 1 | 0% | 1,445 | 2,653 | +84% | 0 | 0 | — |
case-14 | fail→fail | 8,678 | 7,391 | -15% | 1 | 1 | 0% | 1,430 | 2,697 | +89% | 0 | 0 | — |
case-15 | fail→fail | 13,286 | 14,526 | +9% | 1 | 1 | 0% | 1,825 | 3,696 | +103% | 0 | 0 | — |
case-16 | pass→pass | 12,778 | 9,973 | -22% | 1 | 1 | 0% | 1,838 | 3,123 | +70% | 0 | 0 | — |
case-17 | fail→fail | 8,029 | 8,637 | +8% | 1 | 1 | 0% | 1,403 | 3,207 | +129% | 0 | 0 | — |
case-18 | pass→pass | 21,573 | 16,538 | -23% | 1 | 1 | 0% | 3,253 | 4,247 | +31% | 0 | 0 | — |
case-19 | pass→pass | 20,244 | 22,724 | +12% | 1 | 1 | 0% | 4,951 | 6,615 | +34% | 0 | 0 | — |
case-20 | pass→pass | 11,495 | 17,046 | +48% | 1 | 1 | 0% | 1,666 | 3,099 | +86% | 0 | 0 | — |
case-21 | pass→pass | 10,739 | 13,296 | +24% | 1 | 1 | 0% | 1,495 | 3,356 | +124% | 0 | 0 | — |
case-22 | pass→pass | 14,073 | 15,750 | +12% | 1 | 1 | 0% | 2,297 | 3,800 | +65% | 0 | 0 | — |
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 +9 percentage points is the difference between those two pass rates over the 22 comparable cases.
Without the skill loaded, the model failed this case. With it loaded, the same prompt on the same model passed. This is one improved case from the latest verified run; every case, including any that regressed, is in the table above.
Other measured skills in the registry, with their headline benchmark lift.