poemswe/co-researcher

Overview

This skill enforces professional academic writing norms when producing any research prose: literature reviews, syntheses, methods, discussions, abstracts, and other outputs aimed at scholarly audiences. It transforms vague, AI-like text into precise, evidence-centered prose that names studies, methods, samples, and years. The result reads like an experienced researcher wrote it: confident, specific, and disciplinarily appropriate.

How this skill works

The skill inspects drafts for five AI-generated anti-patterns (hedging soup, formulaic transitions, structural monotony, abstraction fog, and voice erasure) and applies corrective edits. It enforces a sequence: identify discipline and audience, draft with specificity, vary paragraph shapes deliberately, use transitions only when they express logical relationships, and run a five-point self-audit before output. The skill also adjusts register and citation integration according to STEM, social sciences, humanities, or interdisciplinary conventions.

When to use it

• Writing or revising any text intended for academic publication or coursework
• Transforming AI-generated or draft prose into publishable-seeming research writing
• Preparing literature reviews, methods sections, results, discussions, or abstracts
• Converting high-level summaries into citation-rich, discipline-aware paragraphs
• Polishing authorial voice to match disciplinary norms

Best practices

• Always specify study details: author, year, sample size, and method rather than using vague phrases
• Use first person where disciplinary norms allow (we in social sciences, I in many humanities) to restore authorial presence
• Limit hedging: fewer than two hedging words per paragraph and replace stacked qualifiers with precise evidence statements
• Delete empty transitions like furthermore/moreover unless they signal an actual logical relation
• Vary paragraph shapes deliberately: mix single-claim, extended-argument, and pivot paragraphs within each section

Example use cases

• Revise a literature review that currently says "many studies" into one that cites four named studies with samples and outcomes
• Edit a methods section to state: "We interviewed 32 nurses across three hospitals in 2021 and analyzed transcripts using Braun and Clarke's (2006) six-phase thematic analysis"
• Convert a discussion draft full of hedges into a concise statement of what the evidence supports and what limits remain
• Prepare an abstract that reports concrete results (effect sizes, confidence intervals, sample) rather than general conclusions
• Adjust register for a multidisciplinary grant proposal so jargon is defined and authorial stance is clear