Claude Novel Workflow — Productivity skill for Claude Code
Canon-driven multi-agent Claude Code workflow for long-form fiction, with a real working novel as the example project.
How to install Claude Novel Workflow
This entry records only its repository, not the path inside it, so there is no
exact command to give. Open MintoTsukino/claude-novel-workflow and copy the folder into
~/.claude/skills/, or the file into ~/.claude/agents/.
What Claude Novel Workflow does
Canon-driven multi-agent Claude Code workflow for long-form fiction, with a real working novel as the example project.
Alternatives in Productivity
- Tradingview MCP — Real-time crypto & stock screening, advanced technical indicators, Bollinger Bands intelligence, candlestick p 2.6k ★
- Repurpose — Workflow recipe — turn one blog post, video, or idea into a full platform-native content pack (thread, LinkedI 1.3k ★
- Claude Code Hooks Multi Agent Observability — Real-time monitoring for multi-agent Claude Code sessions via hook event tracking 1.3k ★
README
claude-novel-workflow
A canon-driven, multi-agent Claude Code workflow for long-form fiction writing — with a real, working project as the reference example.
Built and used by a Japanese light novel author (15+ traditionally published fantasy books, 1M+ views on a previous AI-assisted novel). This repo is the publishable snapshot of the workflow that produced **137,806 characters across 20 chapters in a single 96-minute Claude Code run** — at the author's own writing quality.

What This Is
Three things, in one repo:
- A working example (
example_project/) — A real, in-progress 100-chapter web novel. 70 chapters drafted on the author's machine (~455,000 Japanese characters); the public copy includes the prose, briefs, plots, and scene cards for ep001–ep010, plus the full canon snapshot at ep070 state, one batch review (ep002–ep006), and the full style engine. ep011–ep070 prose is held back for spoiler reasons. - English-language templates (
templates/) — Blank versions of every canon and per-episode file, with guidance comments. Use these to start your own project. - Documentation (
docs/) — Philosophy, architecture, getting started, three-layer review system, how to adapt the example, case study with real numbers.
What This Is Not
- Not a template you copy and fill in. The example project's contents are story-specific. The architecture transfers; the contents don't.
- Not a "press a button, get a novel" tool. It's a structured, iterative workflow that requires you to build canon, exercise judgment, and polish drafts by hand.
- Not a maintained product. This is a published snapshot of a real working project. It will not receive ongoing updates or feature releases. Fork it freely.
Who This Is For
You'll get value from this if you're:
- A novelist (published or aspiring) experimenting seriously with AI-assisted writing.
Related Skills
Claude Workflows
Claude Code plugin: two workflow skills layered on Superpowers. fixing-raw-bugs triages unstructured bug repor
Claude Code Task Shepherd Skill
Claude Code skill that runs a long, multi-segment task to completion by keeping real state in a plan file inst
Multi Agent Training Grpo
A multi-agent system trained with GRPO for reliable long-horizon task planning and execution.
M Orchestrated Dev
Multi-Agent Development Workflow: Research-Driven Dev-Review Cycles
Long Wind
长风 Long Wind — a real-time wuxia action game in the browser (three.js). Every line of code written by Claude O
Pixel Agents
Live pixel art visualization of Claude Code agents working in real time
Related Agents
Deep Fiction Writer
Genre master for purpose-driven and introspective fiction - leads deep narratives while coordinating with the
Lw Implementer
Writes and modifies code. Use for multi-file implementation, refactors, bug fixes, and anything requiring real
Chair Continuity
The continuity chair in the docs-assist authoring loop. Reconciles a packet's example values, naming, and sequ