SuperAPI AI Directory

counterargument

General & other

View repository

Updated 10-16 08:00

About this skill

Counterargument Generation

Generate adversarial critiques to stress-test research claims before writing or submitting.

Perspectives

Perspective Icon Focus
Methodological 🔬 Study design, measurement validity, statistical approach, reproducibility
Theoretical 📚 Conceptual framework, alternative explanations, literature gaps
Practical ⚙️ Feasibility, generalizability, real-world applicability, ethical concerns

Workflow

  1. Identify the core claims — extract the main hypotheses or arguments
  2. Generate 4–6 critiques — at least one per perspective, more for weaker claims
  3. Rate severity — HIGH (fatal flaw), MEDIUM (significant concern), LOW (minor point)
  4. Provide rebuttal strategy — for each critique, suggest how to respond or strengthen the research

Output Format

# ⚔️ Counterargument Analysis

## Summary
[2–3 sentences: overall assessment of the research's vulnerability to critique]

## Counterarguments

### 📚 Theoretical Critique — [SHORT TITLE]

**Challenged Claim:** [Exact claim being challenged]

**Critique:** [Specific, evidence-grounded critique. Reference timescales, measurement limits,
alternative frameworks, or published contradictory evidence where possible.]

**Rebuttal Strategy:** [Concrete suggestion: what experiment, analysis, or argument would
address this critique]

---

### 🔬 Methodological Critique — [SHORT TITLE]
...

Quality criteria

  • Specific: Reference exact claims, timescales, measurements — not vague "this is uncertain"
  • Grounded: Connect to actual limitations in the content, not generic criticism
  • Expert-level: Write as a senior reviewer at a top journal would
  • Constructive: Every critique ends with a path forward (rebuttal strategy)

Examples of strong vs weak critiques

Weak: "The sample size might be too small."

Strong: "The proposed sample of n=15 animals achieves 80% power only for effect sizes of η²≥0.25. Given the typical effect sizes in STDP studies (η²=0.10–0.15, e.g., Bi & Poo 1998), this is underpowered. A power analysis targeting η²=0.12 would require n=38 per condition."