writing-tools

AI-writing humanizer for KO/EN/ZH/JA

Free, open-source AI writing humanizer and detector. 55 patterns, 5 voices, a 0-100 AI-tell score, and nothing leaves your machine.

Clarity - an Agent skill for clearer writing

Agent Skills for end-to-end story writing in markdown, packaged as Codex and Claude Code plugins.

A local-first desktop workspace for worldbuilding, game narrative, quests, maps, relationships, novels, scripts, and AI-assisted writing.

AI协作小说创作框架 - 结构化知识库 + 双入口初始化 + 5阶段工作流

Catches writing that sounds robotic. Shows you why, then rewrites it in a voice that sounds like you.

Claude Code plugin：去除文章中的 AI 寫作痕跡，使文字更自然、更有人味。以台灣正體中文為主。

The digital cork board for filmmakers. Index cards, acts, arcs, and episodes — a local-first planning wall for features, TV, shorts, music videos, commercials, and docs. macOS, Windows & Linux.uggested Donation of $30 if you you can to help me keep making these tools.https://ko-fi.com/samwasserman

Writing workshop in a box. Structured pipeline for generating complete story architecture: novels, screenplays, film or any narrative form. Use manually, or with an LLM to automate the process.

NarrativeNode - A visual plot planner for writers. Built like a node editor , designed for storytellers. With fully optional AI integration via chat and tooluse.

SillyTavern character card & lorebook workspace with editing, AI translation, AI generation, patching, and PNG export.

Clean the AI out of text and detect AI slop - bilingual (EN+RU) Claude skill. Rewrites AI cliches into natural prose, rates AI-likeness, and tells decorative AI emoji from genuine human emoji.

A Macintosh System 6 desktop where AI work stays visible. Local-first, model-agnostic, no framework.

Remove AI writing patterns from Japanese prose. Japanese-localized derivative of hardikpandya/stop-slop (MIT).

Local-first semantic topology visualization for long-form writing, built with Ollama embeddings, Mapper-style graph analysis, and Three.js.

Sam Paech's eqbench Slop Score as an agent skill — flags AI-slop words, trigrams, and 'not X, but Y' patterns in writing. Zero-install, works in Claude Code/Cursor/Codex.

Score a Reddit draft before you post it - eight model judgements, ~120 rewrites built from your own sentences, word-level diffs. One file, zero dependencies, runs on your machine.

A local, in-browser slop meter for your own writing. Grades drafts from Human to Slop 9000 and highlights the cliches to cut. No network, no tracking, ~14KB of readable JS.