Creative Writing Skills

Write novels, short stories, and serial fiction with AI that maintains your voice, tracks your continuity, and gets better the more you use it. From first brainstorm to polished draft — specialized agents handle each mode of work (writing, critiquing, revising, exploring) while shared skills carry the craft methodology.

What you get:

Brainstorm without committing — explore plot options, character arcs, and world mechanics with multiple AI perspectives before deciding anything
Write in your voice — create style files from your existing prose, then draft new scenes that match
Catch your own mistakes — structured critique, continuity checks, and simulated reader reactions
Keep everything in sync — knowledge base updates as your story evolves

Quick Start

meridian mars add haowjy/creative-writing-skills
meridian mars sync
meridian bootstrap

Installation

Mars (Meridian)

meridian mars add haowjy/creative-writing-skills
meridian mars sync

Then run meridian to start a session with muse. First-time setup runs the bootstrap automatically.

Claude Code / Cowork

Claude Code and Cowork use the same plugin format.

Claude Code — add the marketplace and install:

/plugin marketplace add haowjy/creative-writing-skills
/plugin install creative-writing-skills@cw

Cowork — open the Claude Desktop app, go to Customize → "+" → Create plugin → enter haowjy/creative-writing-skills or upload from a local clone.

Once installed, start a session with muse as your agent:

claude --agent creative-writing-skills:muse

Run the one-time project setup to create your CLAUDE.md and kb/ structure:

/creative-writing-skills:project-setup

Claude.ai (Skills Only)

Claude.ai supports skills but not agents — you get the craft methodology without multi-agent orchestration.

Clone this repo and run python scripts/create_skill_zips.py to generate .skill files in zips/
Go to claude.ai → Customize → Skills → "+" → Upload skill
Upload each .skill file

Recommended starting set: writing-principles, prose-writing, scene-construction, prose-critique.

How It Works

flowchart TB
    You([You]) --> M[muse]

    M --> Explore
    M --> Draft
    M --> Maintain

    subgraph Explore ["Explore & Plan"]
        direction LR
        B[brainstormer] ~~~ CS[character-sim]
        CS ~~~ O[outliner]
    end

    subgraph Draft ["Draft & Revise"]
        direction TB
        W[writer] --> CR[critic]
        CR -->|synthesis| RW[revision-writer]
        RW -.->|iterate| CR
        M -.->|pivotal scenes| RS[reader-sim]
        M -.->|voice check| CC[continuity-checker]
    end

    subgraph Maintain ["Knowledge"]
        direction LR
        CH[chronicler]
        SC[style-creator]
    end

    Explore -->|direction confirmed| Draft
    Draft -->|decisions & facts| Maintain
    Maintain -.->|context for next session| Explore

Explore: Fan out brainstormers for creative variety. Spawn character-sims to discover voices. Use outliners to shape structure once a direction is chosen.

Draft & Revise: Muse runs the write/critique/revise loop — writer produces prose, critics evaluate across the four reward channels (transportation, aesthetic, social simulation, flow), revision-writer fixes specific issues. Reader-sim gives experiential signal on pivotal scenes.

Knowledge: Chronicler extracts facts from completed chapters into the kb. Style-creator captures voice patterns from prose samples. The kb grows as the project evolves, giving every future agent accurate context.

Agents

Agent	Role
muse	Author's creative partner — intent capture, brainstorming, drafting loops, knowledge maintenance
writer	Generative prose from scene briefs in the project's voice
revision-writer	Surgical revision from critique — preserves voice, fixes specific issues
bridge-writer	Transitions, time compression, connective passages between pivotal scenes
critic	Adversarial critique across the four reader reward channels
reader-sim	Simulates a reader's experience, reports moment-by-moment
character-sim	In-character voice performance for discovery and relationship testing
continuity-checker	Checks drafts against established canon
brainstormer	Wide-open option generation on a scoped question
outliner	Structural decomposition into beat sheets and arc maps
style-creator	Analyzes existing prose to create style reference files
chronicler	Extracts story facts from chapters into the kb

Skills

Skill	Purpose
writing-principles	Four reward channels, AI failure modes, reader psychology
prose-writing	Immersion patterns — psychic distance, rhythm, sensory grounding, interiority
scene-construction	Beat-level craft — scene entry, dialogue, pacing, transitions
prose-critique	Adversarial reading methodology with focus areas
style-analysis	How to analyze prose and produce style reference files
story-architecture	Arc shape, tension curves, structural analysis
story-context	Context scoping — what to pass to each agent type
brainstorming	Exploratory capture with source tagging
kb-management	Maintaining the story knowledge base — page conventions, organization, when to create vs update
writing-artifacts	File conventions — kb/ for durable knowledge, work/ for scratch
writing-issues	Issue tracking across revision cycles
writing-staffing	Team composition — which agents, how many, what focus areas
project-setup	One-time guided setup — interviews about the project, creates CLAUDE.md and kb structure

Project Layout

my-story/
├── CLAUDE.md              # Project conventions (created by project-setup)
├── story/                 # Chapters and manuscript
├── work/                  # Current drafting effort
│   ├── outline/
│   ├── drafts/
│   ├── critique-reports/
│   └── brainstorm/
└── kb/                    # Durable knowledge base
    ├── styles/            # Voice reference files
    ├── characters/        # Character state and profiles
    ├── world/             # Locations, lore, systems
    ├── timeline/          # Chronology
    ├── canon/             # Established facts
    └── issues/            # Tracked writing problems

Compatibility

Feature	Claude Code	Cowork	Mars (Meridian)	Claude.ai
All agents	Yes (flat)	Yes (flat)	Yes (hierarchical)	No
All skills	Yes	Yes	Yes	Upload as zip
Multi-agent orchestration	Via muse	Via muse	Via muse → bard → workers	No
Project setup	Yes	Yes	Yes	No

Claude Code and Cowork use the same plugin format with all agents as flat subagents under muse. The Meridian version adds intermediate orchestrators (bard for drafting, lore-keeper for knowledge maintenance) for deeper context isolation. Claude.ai supports skills only — upload individual skill zips for craft methodology without orchestration.

Current Experiments

Rhetorical questions in skill prompts. The economy section in writing-principles uses rhetorical questions ("what can you leave out and still have the scene work?") rather than declarative statements. LLMs can distinguish rhetorical from information-seeking questions internally (arxiv 2604.14128), and Self-Ask prompting shows questions improve reasoning — but no research directly tests whether rhetorical questions in system prompts improve task performance vs. equivalent declaratives. Keeping the rhetorical form to see if it activates a self-check loop that declaratives don't.

Development

Validate package

meridian mars check

Release

mars version patch              # bump, commit, tag
mars version patch --push       # bump, commit, tag, push

License

Apache License 2.0. See LICENSE.