Skip to content

README.md

Latest commit

 

History

History
175 lines (117 loc) · 5.77 KB

README.md

File metadata and controls

175 lines (117 loc) · 5.77 KB
Raw

content-ops

Work in progress. The plugin works — but it's still being polished. Some defaults are borrowed from the project it was built on and aren't fully agnostic yet. Expect rough edges and active iteration. Follow progress on LinkedIn.

A Claude Code plugin that turns your content strategy into a repeatable writing system — so you can publish faster without sacrificing quality.

You define your content pillars, goals, and style. content-ops handles research, writing, translation, internal linking, fact-checking, and indexing. You review, approve, and ship.

Works with any markdown-based static site (Astro, Next.js with Contentlayer, Hugo, and others) — if your content has frontmatter and a body, it should work.

Part of the content-stack marketplace.

What it does

Instead of writing each article from scratch, you run a single command and get a research-backed, style-reviewed, internally-linked draft — ready for your review.

The typical workflow:

  • Add ideas to a backlog as they come to you
  • When ready, ask content-ops to pick from the backlog — or kick off an interactive session for something specific
  • Claude researches, drafts, checks facts, enforces your style, links to related content, and updates your trackers
  • You review the PR, iterate if needed, and ship

It's the same loop you'd run with a copywriter — but always available, always on-brief.

Quick start

1. Install

Install via the Claude Code plugin marketplace:

/plugin marketplace add pcamarajr/content-stack
/plugin install content-ops@content-stack

2. Init

Run the setup wizard inside Claude Code:

/init

This walks you through seven setup rounds: your author info and languages, content types and paths, style guide and reference articles, content strategy and pillars, infrastructure (backlog, trackers), image generation, and link building strategy.

You can run rounds individually:

/init project
/init content-types
/init style
/init strategy
/init infra
/init images
/init link-building

3. Index your content

/reindex

Scans your existing content and builds .content-ops/content-index.json — the knowledge layer the agents use for linking and discovery.

4. Write something

/write-content article "Getting started with Docker"
/write-content backlog 3
/write-content

That's it. Claude takes it from there.

If you want a ready Astro base to test this quickly, start from:

How it works

content-ops is a phase-based orchestrator. When you run /write-content, it coordinates a pipeline of specialized agents — each focused on one job.

flowchart TD
    User([You]) -->|/write-content| WC[write-content skill]

    WC --> P1[Phase 1\nParse arguments]
    P1 --> P2[Phase 2\nLoad config + guidelines]
    P2 --> P3[Phase 3\nPlan topic]
    P3 --> P4[Phase 4\nResearch]
    P4 --> P5[Phase 5\nDraft]
    P5 --> P55[Phase 5.5\nGenerate images\noptional]
    P55 --> P6[Phase 6\nStyle review]
    P6 --> P7[Phase 7\nGlossary check]
    P7 --> P8[Phase 8\nBidirectional linking]
    P8 --> P9[Phase 9\nReindex + commit]

    P4 <-->|cache hit/miss| RC[(Research cache\n.content-ops/research-cache/)]
    P5 --> DW[draft-writer agent]
    P4 --> CR[content-researcher agent]
    P6 --> SE[style-enforcer agent]
    P7 --> GC[glossary-creator agent]
    P8 --> CL[content-linker agent]
    CL <-->|reads| CI[(Content index\n.content-ops/content-index.json)]

    P9 --> Commit([git commit])
Loading

All knowledge lives in files — no embeddings, no vector DBs, no external APIs required for the core pipeline.

→ Full technical walkthrough

Skills at a glance

Skill What you type What happens
/init /init or /init [round] Setup wizard — creates config, guides, and trackers
/write-content /write-content article "Topic" Full pipeline: research → draft → style → link → commit
/translate /translate es Localize content to a target language
/fact-check /fact-check path/to/article.md Verify every claim against trusted sources
/review-content /review-content path/to/article.md Style, tone, structure, and linking audit
/suggest-content /suggest-content 5 Suggest next articles from gaps in your strategy
/reindex /reindex Rebuild the content index from current files

→ Full skills reference

Requirements

  • Claude Code CLI
  • Node.js 22+
  • A markdown-based static site with frontmatter content

No API keys needed for the core workflow. Image generation is optional and uses Google Gemini or OpenAI when enabled.

Documentation
How it works Architecture, agent pipeline, phase breakdown
Configuration Full config schema and .content-ops/ layout
Skills Every skill with examples and options
Agents What each agent does and when it runs
Knowledge layer Content index, research cache, and how linking works

Status

This plugin is actively being developed. Current state:

  • Works: init wizard, write-content pipeline, translation, fact-check, review, suggest, reindex, internal linking, research cache
  • In progress: removing project-specific assumptions to make it fully framework-agnostic
  • Not yet: automation / CI mode, public examples, full test coverage

Feedback and issues welcome on GitHub.

License

MIT — by Pedro Camara Jr