Browse / Content & marketing / content-studio
content-studio
tsgx1990
AI content studio for Claude Code: research → draft → review → publish, enforced by zero-dependency deterministic gates.
Open tsgx1990/content-studio → On the boundary of the four rules.
Works with any agent
ships AGENTS.md, the cross-tool instruction file.
In its own words
Skills + editorial agents for SEO, long-form, kids', YouTube/Xiaohongshu/short-drama, and more.
A Claude Code plugin that turns content production into a disciplined pipeline: research → draft → review → publish, with deterministic, zero-dependency gates that refuse to ship keyword-stuffed, fabricated, or non-compliant content.
It is Claude Code native — the workflow engine is the agent itself following a set of skills; the only code is small node gate scripts (no build step, no npm install, no dependencies).
What you get
- Content-type skills — SEO articles, long-form fiction, children's stories, YouTube scripts,
Xiaohongshu (小红书) notes, short dramas (微短剧), graded readers, interactive/game/audio stories.
- Editorial subagents — a chief editor, SEO editor, copyright-risk reviewer, a (mandatory,
blocking) child-content-safety editor, and long-form continuity/architecture reviewers.
- Deterministic gates — one per content type plus shared publish, AI-tell, freshness, and
Chinese-marketing-compliance scanners. The gates, not trust, are the source of truth.
Install
Then, in any content workspace, the skills are available (e.g. /keyword-research, /seo-article, /content-review, /publish-content). Skills are namespaced by the plugin, e.g. /content-studio:seo-article.
Updating: bump the plugin and the marketplace, push, and users run
/plugin update. Pinning aversionin the manifest means users only receive updates when you bump it.
The pipeline (mandatory)
No content reaches a publish target without going through:
The repository's own description, then 48% of its README. Read the rest on GitHub ↗
Read from the repository's git tree on 2026-08-17.