seandavi

Agentic Writing Patterns — AI skill for Claude Code

AI community

Design patterns of agentic AI in scholarly writing — narrative-review paper + lit review wiki.

How to install Agentic Writing Patterns

This entry records only its repository, not the path inside it, so there is no exact command to give. Open seandavi/agentic-writing-patterns and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Agentic Writing Patterns does

Design patterns of agentic AI in scholarly writing — narrative-review paper + lit review wiki. PRISMA-trAIce-compliant; senior author Sean Davis, Claude (Anthropic) as postdoc-equivalent collaborator.

Alternatives in AI

  • Humanizer: Remove AI Writing Patterns — You are a writing editor that identifies and removes signs of AI-generated text to make writing sound more nat 10.5k ★
  • Claudian — An Obsidian plugin that embeds Claude Code as an AI collaborator in your vault 7.7k ★
  • Welcome — AI Research Skills — You now have access to 86 production-ready skills covering the entire AI research lifecycle: literature survey 5.4k ★

README

agentic-writing-patterns

A design-patterns review of agentic AI in scholarly writing — catalog, evidence base, and methodology.

This repo is both the working environment for a narrative review paper and the methodology evidence the paper will cite. The wiki is the synthesis layer; issues track decisions and progress; the eventual manuscript will live in `paper/`.


Scope

**In scope**

  • Design patterns for AI-assisted scholarly writing in the prompted-LLM era (2023+)
  • Patterns identified across the writing pipeline: topical research, gap analysis, drafting, citation handling, revision, reviewer simulation
  • Patterns extracted from published, citable sources (DOI / arXiv / PMID resolvable)

**Out of scope**

  • General LLM benchmarks unrelated to scholarly writing
  • Non-scholarly writing tools (creative, journalistic, marketing)
  • Finetuned-LLM workflows (largely superseded by prompted-model approaches; see Susnjak 2023)
  • Implementation of the patterns (see scriptorium for one downstream instance, scoped to manuscript-stage review)

Methodology

PRISMA-trAIce-compliant systematic review ([Holst et al. 2025](wiki/sources/holst-2025-prisma-traice.md)). Full protocol in [`wiki/methodology.md`](wiki/methodology.md).

Operational anchors:

  • Search transparency — every query, tool version, and date logged in wiki/log.md
  • Two-phase triage — title/abstract sweep, then full-source ingest
  • Pattern-evidence threshold — concept pages require ≥3 sources before promotion from provisional
  • Human-as-Quality-Gate — Sean reviews every source-page summary before it informs a concept page

Authorship and AI collaboration

This work uses Claude (Anthropic) as a research and writing collaborator under continuous human supervision.

  • Sean Davis — senior author and intellectual director. Sets scope, makes judgment calls on what counts as a pattern, g