Agent-Skills.md

Agent Skills: IMRaD Reviewer

Use when reviewing an IMRaD draft for structural integrity, evidentiary support, and section-boundary correctness.

UncategorizedID: narumiruna/agent-skills/imrad-reviewer

Repository

narumiruna/agent-skills
narumirunaLicense: MIT
6

Install this agent skill to your local

pnpm dlx add-skill https://github.com/narumiruna/agent-skills/tree/HEAD/skills/imrad-reviewer

Skill Files

Browse the full folder contents for imrad-reviewer.

Download Skill

Loading file tree…

skills/imrad-reviewer/SKILL.md

Skill Metadata

Name
imrad-reviewer
Description
Use when reviewing an IMRaD draft for structural integrity, evidentiary support, and section-boundary correctness.

IMRaD Reviewer

Purpose

Review an existing IMRaD draft for structural, logical, and evidentiary quality.

When to use

Use this skill when the user asks to:

  • Audit IMRaD structure quality
  • Find section-boundary violations
  • Identify unsupported or over-interpreted claims
  • Decide whether a draft is structurally publishable

When NOT to use

Do NOT use this skill when:

  • The user asks for first-pass drafting from raw notes
  • The user asks for full document rewrite by default
  • The input is not an IMRaD draft and no review target exists

Review Criteria

  • Introduction defines a concrete question/problem
  • Methods are sufficient for the document scope
  • Results are objective and non-interpretive
  • Discussion interprets only what Results support
  • Linkage is intact: question -> method -> result -> interpretation
  • Limitations and assumptions are present where needed

Review Workflow

  1. Validate section presence and order.
  2. Check section-boundary purity.
  3. Trace major claims to Results evidence or explicit assumptions.
  4. Flag unsupported claims and over-interpretation.
  5. Produce actionable fixes with severity.

Output Requirements

The output MUST include:

  • Overall assessment
  • Strengths
  • Issues list, where each issue includes:
    • severity: high, medium, or low
    • section
    • problem
    • why it matters
    • suggested fix
  • Optional final verdict:
    • structurally sound
    • needs revision
    • unsuitable as IMRaD

Severity Model

  • High: structural breakage, fabricated or unsupported core claims, major section mixing
  • Medium: weak linkage, under-specified methods, incomplete limitations
  • Low: clarity, concision, or minor organization issues

Truthfulness Constraints

  • The reviewer MUST NOT invent missing evidence
  • The reviewer MUST distinguish empirical vs non-empirical results
  • Non-empirical results MUST be explicitly labeled as inferred, hypothetical, expected, or simulated

Failure Modes

  • Rewriting the full draft instead of reviewing
  • Mixing structural critique with factual speculation
  • Missing contradictions between Results and Discussion
  • Providing non-actionable feedback