QN

qiyan233/novelops-skill

Developer tools

This page was translated automatically. Technical details may differ from the original. View the original

33 stars Quality 59 Trend 20MIT

A long-form novel workflow skill framework for OpenClaw, supporting state management, continuity auditing, and next-chapter context construction

Overview

A ** long novel workflow skill. It turns long novels / serialized / web novels / fan fiction into a ** — the focus is not on "writing one chapter at a time," but on maintaining the settings and retrievable state by chapter 50. The underlying layer is a pure Python 3 CLI plus plain text state files, without binding to a specific host: OpenClaw, Claude Code, agents within the IDE and terminals can all drive it. See details on how to use it in different agent applications. Inspiration reference: https://github.com/Narcooo/inkos。本项目是受其启发的 skill skeleton, not the official port of the original project. The real difficulty of long-form writing isn't "not being able to write the next chapter," but the **. This warehouse breaks down the matter into executable processes: - Maintain worldview, current situation, and role recognition boundaries - Organize "What to write in the next chapter, what to avoid" into a structured work package - Delegate to LLM drafting (Hermes priority, default is local Ollama) - Audit and revision closed loop - Write newly occurring facts back to status - Revert and compare at any time - Want to write novels long-term, multi-chapter, and retrospectively, need to maintain chapter summaries, character status, foreshadowing, and current situations - Want a novel writing skill similar to InkOS - Hope for auditing, revision, status updates, snapshots, and other steps beyond writing **Please use python scripts/novelops_cli.py ...; scripts/*.sh is a legacy entry point and requires bash. All functions have equivalent Python commands. The draft can be written by a human or agent, or directly given to the LLM; revise Internally links knowledge-check / audit / revision-plan / spot-fixes. For command details, see the CLI entry.

Overview (machine translated)

Current Version: 1.1.0

A Agent Application Long Novel Workflow Skill. It combines long novels, serializations, web novels, and fan writing Run into a long-term maintainable process—the focus is not on “writing one chapter at a time,” but on ensuring that by chapter 50, the settings are stable and the state is traceable.

The underlying layer is a pure Python 3 CLI plus plain text status files, without binding specific hosts: OpenClaw, Claude Code, Agents within the IDE and terminals can drive it; see its use in different agent applications for details.

Inspiration reference InkOS: . This project is inspired by the skill skeleton, Not an official port of the original project.

Chinese | English


What problems does it solve?

The real difficulty in long-form writing isn’t “not being able to write the next chapter,” but remembering who learned what in chapter 3 by the time you reach chapter 50. This warehouse breaks down the matter into executable processes:

  • truth files Maintain worldview, current situations, and character cognitive boundaries
  • write-next Organize “What to write in the next chapter, what to avoid” into a structured work package
  • draft Drafted by LLM (Hermes priority, default local Ollama)
  • revise / auto-revise Auditing and revision closed loop
  • extract-state / state-update Write the newly occurred fact back to the state
  • snapshot / diff Revert and compare at any time

It is not just responsible for “writing,” but organizing multiple chapters of writing.

Who is it for?

Suitable for:

  • If you want to write a novel long-term, multi-chapter, and retrospective, you need to maintain chapter summaries, character status, foreshadowing, and current situations
  • Want to build a novel writing skill similar to InkOS
  • I hope that beyond writing, there will be audits, revisions, status updates, snapshots, and other steps

Not very suitable: I just want to create a short article on the spot. This repository will be noticeably heavy.

Get Started Quickly

Requires Python 3, no third-party dependencies.

Windows / No Bash Environment Please use python scripts/novelops_cli.py ...; scripts/*.sh is a legacy entry point and requires bash. All features have equivalent Python commands.

For more detailed steps, see Installation & Environment and Quick Hands-On Practice.

Core Main Storyline

draft Can be written by a human or agent, or directly handed over to the LLM; revise Threaded internally knowledge-check / audit / revision-plan / spot-fixes。 For command details, see the CLI entry.

Current Capabilities

Already covered:

  • Next Chapter Preparation (write-next, including chapter contracts and scene beats)

This page was translated automatically. Technical details may differ from the original.

Recommended Tools

Try a different keyword or remove a filter.

Install

npx skillfish add qiyan233/novelops-skill
New to MCP? Start with the mcpskill.org guide →