Phrasebank-led academic writing, Q1 readiness, and offline review methodology

A modular writing system for research drafting, synthesis, and review.

This repository turns Academic Phrasebank-style rhetorical guidance into a task-first agent system for manuscript drafting, literature analysis, reviewer response work, and SLR or SMS methodology support. The structure is local, modular, and built to route real academic tasks instead of serving as a flat prompt template.

Read the README Try a starting prompt

How this works

The system is task-first. A natural request determines the section, style, discipline overlay, methodology layer, quality checks, and anti-AI controls.

scholarly-research-writing-review
# Natural request
user: Use $scholarly-research-writing-review.
Write an Introduction for an Elsevier article in engineering.
Topic: explainable AI for medical image diagnosis. Use British English.
# Internal routing
section: introduction
style: elsevier
discipline: engineering
quality: q1_wos_readiness + self_review_protocol
controls: anti_ai + gap_validation
# Result
A rhetorically structured draft with explicit gap framing, controlled novelty claims, and section-level academic language that aligns with the repository modules.
22
Phrase Libraries
10
Methodology Source Notes
5
Journal Style Overlays
4
Core Analysis Workflows

Start from a real prompt

Pick a common task. The prompt below is copied directly from the patterns supported in the README and codebase.

More than a phrase template

The repository combines rhetorical language support with routing rules, standards, workflows, methodology notes, and anti-formulaic controls.

IF THIS WERE ONLY A PROMPT PACK
  • Section phrases would exist without routing logic or task classification.
  • Systematic review work would depend on generic web-style prompting rather than local methodology notes.
  • Novelty framing, Q1 readiness, and self-review would be left implicit.
  • Outputs would drift toward repetitive, polished, AI-sounding prose.
WHAT THIS REPOSITORY ADDS

The phrase libraries support the system. They are not the whole system.

Repository modules

Each card maps to a folder in the repository. Together they cover the full range of writing, review, methodology, and analysis support.

Phrase Libraries

  • phrases/introduction.yml
  • phrases/literature_review.yml
  • phrases/methods.yml
  • phrases/discussion.yml
  • phrases/cautious_language.yml

Rhetorical Moves

  • rhetorical_moves/cars_model.md
  • rhetorical_moves/gap_identification.md
  • rhetorical_moves/novelty_positioning.md
  • rhetorical_moves/methodological_justification.md
  • rhetorical_moves/contribution_framing.md

Quality Standards

Methodology

Runtime & Routing

Skill Package

Bibliometric Analysis

Discipline Overlays

Reviewer Response

Writing Notes

Journal Styles

Workflows

Common questions

No. The README is explicit that this repository is not an official Academic Phrasebank product. It is a derivative, modular scholarly writing and review system that uses the Phrasebank as a primary rhetorical foundation and extends it with routing, methodology, standards, and review workflows.
The codebase and README cover manuscript drafting, literature review writing, gap identification, reviewer response work, single-paper analysis, multi-paper synthesis, manuscript diagnosis, and SLR, SMS, or tertiary-study methodology support.
The repository intentionally keeps Kitchenham, SEGRESS, and PRISMA support as local Markdown or YAML notes under methodology/source_notes/. The goal is to reduce unnecessary web dependence and make review-method tasks reproducible from the local knowledge base first.
It means the default quality bar is not just fluent prose. The system is expected to apply explicit standards for contribution framing, methodological defensibility, rhetorical coherence, and self-review, primarily through the files in standards/ and the gap-validation layer.
The recommended pattern in the README is natural-language prompting, optionally with explicit invocation such as Use $scholarly-research-writing-review. Optional routing labels exist in COMMANDS.md, but they are presented as conventions rather than guaranteed native commands.
The packaged skill under packages/scholarly-research-writing-review/ includes a SKILL.md, agents/openai.yaml, and a bundled references/ tree copied from the repository.
The anti-AI layer is explicit. Files in anti_ai/ target cadence repetition, excessive transition chaining, rhythm symmetry, and over-smoothed paragraph endings so the output stays formal without sounding templated.

Open the right entry points first

These are the files that define how the repository behaves in practice.

CORE RUNTIME
AGENTS.md + runtime/system_prompt.md + runtime/router.md

Use these to understand the system contract, runtime behavior, and routing assumptions.

PACKAGED SKILL
packages/scholarly-research-writing-review/

Use this when you want the installable Codex skill artifact rather than only the source repository.

Also useful: AUTO_ROUTING_RULES.md, TEST_CASES.md, and METHODOLOGY_SOURCES.md.

Open the Skill Package