# Content library for written content: what to keep and where · ScriptGrain

> A content library for writing keeps more than files: the pieces, the drafts and their versions, the briefs, and what your writing has already said. What each kind of tool keeps (CMS, docs, content platforms, voice libraries), and how to set one up that your AI can use.

Canonical: https://scriptgrain.com/reference/written-content-library

# A content library for written content: what to keep, and where it should live

*By Jack Stovell [X](https://x.com/StovBuilds) · [Instagram](https://www.instagram.com/stovbuilds) · [LinkedIn](https://www.linkedin.com/in/jackstovell/) · published 2026-10-10 · checked 2026-10-02*

A written content library is the one place your finished pieces, drafts and briefs live, searchable, with a record of what each piece said. If you write with AI, it also has to hold your voice, so new drafts sound like you and don't repeat you. Most people need more than one tool to do all of that.

## The short answer

Keep five things: finished pieces, drafts with their versions, briefs, a record of what you've already said, and your voice. That's the whole list.

The part people trip over is that the tool you publish from and the tool you keep your writing in are often different tools. A CMS is very good at putting a post on a website. It was built for that job. It wasn't necessarily built to answer "have I told this story before?" or "does this draft sound like me?" So you end up with a publishing tool, a place where drafts get written, and a library that holds the lot.

To be fair, plenty of solo writers get by with a folder of documents and a decent memory. That works until the folder hits a few hundred files and the memory starts to fail. Years of posts, emails and half-finished drafts scattered across docs, a CMS and chat histories is the normal state of things, not a sign you've been careless.

And once AI drafts some of the writing, the stakes go up. A model has no idea what you published last spring. It will happily write the same story again, in a voice that's close to yours but not quite.

## What a written content library should hold

Take each item in turn, in plain words.

Finished pieces come first. These are the posts, newsletters and emails that went out. They're your record of what you actually said, and they're the raw material your voice is measured from. If they live only inside a CMS, you can't easily search them as writing, only as web pages.

Drafts with their versions come next. Every piece of writing changes shape on the way to done. The paragraph you cut on Tuesday is sometimes the one you want back on Friday. A library that keeps the text each edit replaced lets you undo a bad rewrite without panic. That matters more once AI is in the loop, because a redraft can quietly overwrite a version you liked.

Briefs are the third item, and the one most people forget. A brief is the topic, the angle, the audience, the register, the key points and the length. If you write a weekly newsletter, the brief is mostly the same every time. Saving it means you stop retyping it, and a team can start from the same one.

Then there's a record of what has been said. This is the memory. Picture a founder who posts weekly. By month fourteen they cannot remember which stories already went out, so the origin story gets told a third time and the customer anecdote a second. A record of ideas, claims, stories and frameworks, each tied to the words it came from, catches that before the draft does.

Last is the voice itself. It's the pattern across your finished pieces: sentence length, the words you reach for, how you open and close. If you write with AI, the voice is what keeps a draft from sounding like everyone else's. Without it, you're editing generic prose into something that sounds like you, every single time.

## CMS, docs, content platform or voice library

Four kinds of tool come up when people ask where their writing should live, and they do different jobs. The table below sets them side by side; here's how to read it.

A CMS such as WordPress, Contentful or Webflow stores and publishes web pages and posts. If your goal is a live site, it's the right place for the final version of a piece, and nothing else on this list replaces it for that. Its strength is publishing, and that's where it earns its keep.

A docs or notes workspace such as Google Docs or Notion holds documents and lets people edit and comment together. It's where most drafts get written, and it's very good for collaboration. Plenty of people treat it as their library by default, and for a small body of work that's a sensible choice.

A content marketing platform such as StoryChief, HubSpot's Content Hub or CoSchedule plans, routes for approval and schedules content across channels. If you run a calendar with several people and several channels, this is the category built for it. It's about coordination as much as storage.

A voice library, which is what ScriptGrain is, keeps the writing a voice is measured from, the drafts written in it with their scores and versions, and a memory of what that writing has already said. Its job is narrower than the others: voice and memory.

There's a fifth kind worth a passing mention, the digital asset manager. It's built mainly for images, video and brand files, so it solves a different problem from written content.

So which one do you need? Often two or three. Many teams write in docs, publish from a CMS, plan in a content platform and keep the voice and the memory in a voice library. Each tool does the job it was built for.

Kind of tool · Examples · Built for · Keeps the voice?
CMS · WordPress, Contentful, Webflow · Storing and publishing pages and posts · As published text
Docs or notes workspace · Google Docs, Notion · Writing and editing documents together · As documents and any style guide you write
Content marketing platform · StoryChief, HubSpot Content Hub, CoSchedule · Planning, approval and scheduling across channels · As text and any brand settings it offers
Voice library · ScriptGrain · The writing a voice is measured from, drafts with scores and versions, and a memory of what was said · As a measured profile of 45 attributes, read by API and MCP

## How ScriptGrain keeps written content

Here's what ScriptGrain holds, exactly as the app works.

The Library is where every piece of writing you add lives. You can paste text, upload files, import a page by URL (URL import is on paid plans), or save a finished draft from the draft screen with "Save to library". Filter chips split it into All, Pasted, Files, URLs and From drafts. You can search by name or content and sort by newest, oldest, longest or name. Each piece shows which voice profiles were built from it. You can open it, copy it, download it as a .txt file, rename it, delete it, or use it in a new voice. A voice profile takes up to 20 samples.

Drafts live on the Drafts page, and each generated draft carries its voice match score. Version history is the safety net. Whenever a finished draft's text is replaced, whether by your own edit, a paragraph rewrite, a polish or humanise pass, or a change over the API, the text it replaced is kept. The newest 50 versions per draft are stored, and any of them can be restored. The restore is itself kept as a version, so you can't lose your way by going backwards.

Brief templates let you save a brief (topic, angle, audience, register, key points, length) as a named template and start new drafts from it. On a team plan, a template can be shared with the team.

Voice Brain is the memory. It holds the ideas, claims, stories and frameworks your writing has already covered, each kept with the exact words it came from. A novelty check scores a new brief against it as repeats, builds on, or new. It proposes what to write next, and it flags two claims from different pieces that pull against each other. A memory map lets you add, reword, move and delete single memories, and one brain can serve several voices. It's on every plan. You can read more in [Voice Brain](https://scriptgrain.com/voice-brain), and how the system picks up your corrections is covered in [how it learns from your edits](https://scriptgrain.com/learns-from-your-edits).

The voice monitor, on paid plans, scores your published pages against your profile once a month. Writer and Operator cover 5 URLs, Studio 15 and Agency 50. It's a check that your live writing still sounds like the profile.

Teams share the lot. Studio (3 seats included) and Agency (10 seats) share voices, library pieces, drafts and templates with the team, and private items stay private.

## Using the library from ChatGPT or Claude

You might not want to leave the assistant you already use. Fair enough. The profile, the memory, the drafts and the templates are reachable from ChatGPT, Claude and any MCP client, over the API and the MCP server (41 tools). They're read live each time. They aren't a pasted prompt.

That distinction is the point. A pasted prompt is a snapshot: it goes stale the moment you add a new piece or correct a memory. Reading live means the assistant sees the current profile and the current memory every time it asks. Add a finished post this afternoon, and the next draft already knows about it.

In practice, you could ask Claude for a draft and, with the MCP server connected, have it run the novelty check against your memory before it writes a word. The Free plan includes the API and MCP server, so you can try this without paying. Details are on the [API and MCP page](https://scriptgrain.com/ai-writing-api-and-mcp-server).

## Keeping a library useful

A library only helps if you tend it. Four habits do most of the work.

Name pieces so search finds them. "Newsletter 14" tells you nothing in a year. "Newsletter: why we stopped discounting" tells you what it said. Since search runs on name and content, a clear name is cheap insurance, and Library has no folders or tags to lean on.

Keep only writing that's really yours, or writing you may use, in a voice. A profile built from a blend of your writing and someone else's will drift towards the average of both. If a guest post or a client's copy slips in, take it out. Who signs off which check, consent to use someone else's words included, is set out on the [content governance page](https://scriptgrain.com/reference/content-governance-for-written-content).

Add finished pieces back so the memory stays current. This is the one people skip. You publish, you move on, and three months later the library is a quarter behind what you've actually said. Saving each finished draft to the library takes seconds, and it keeps the novelty check honest.

Retire what no longer represents you. An old post written before you changed your mind, or in a register you've outgrown, drags the profile backwards. Delete it, or reword the memory that came from it.

If you want a free way to see what your existing pieces already say, the [content ideas generator](https://scriptgrain.com/tools/content-ideas-generator) reads them and shows you. And if several writers share one voice, the [consistency checker](https://scriptgrain.com/tools/brand-voice-consistency-checker) checks their drafts against it.

## When ScriptGrain is the wrong place

ScriptGrain is not a CMS. It doesn't publish to a website, schedule posts, run a content calendar, or route drafts through approval steps. It has no folders or tags in the Library. You copy or download the text into wherever it's published.

So if your team needs a calendar, approval routing or scheduling, keep those in your publishing tool or content platform. That's what they're built for. Then use ScriptGrain for the two things it does: the voice and the memory.

On price, the Free plan gives you one voice profile and one writing analysis covering all 45 attributes, plus the API and MCP server, and one free draft for the life of the account. Generation is otherwise paid, and the Writer plan is £12 a month.

## Limits

This page cannot choose a CMS or a content platform for you. It compares kinds of tool, not products, and it does not cover images, video or brand files.

ScriptGrain's Library holds writing for voices and drafts. It has no folders or tags, and it does not publish, schedule or route approvals, so a team that needs those still needs a publishing tool.

A voice profile is only as good as the writing behind it. A library full of pieces someone else wrote, or old work that no longer sounds like you, gives a profile that drifts from you.

## Questions

### What is a content library?

A content library is the one place your written work lives so you can find and reuse it. For writing, that means finished pieces, drafts and their versions, briefs, and a record of what each piece said. If you draft with AI, it should also hold your voice, so new drafts match how you write.

### What is the difference between a content library and a CMS?

A CMS stores and publishes web pages and posts; its job is the live site. A content library keeps your writing as writing: searchable, with drafts, versions, briefs and a record of what you've said. Many people use both, publishing from the CMS and keeping the working record somewhere built for that.

### Where should I keep all my writing?

Keep finished pieces and drafts somewhere searchable that you control, and publish from a separate tool if you need a website. A docs workspace works for a small body of writing. Once you have years of it, or you draft with AI, a library that tracks voice and what you've already said earns its place.

### Can AI use my content library?

Yes, if the library is reachable from the assistant. Over an API or MCP server, ChatGPT or Claude can read your profile, memory, drafts and templates live, rather than from a pasted prompt. That way the draft reflects your current writing, and the assistant can check what you've already covered before it starts.

### How do I stop AI repeating what I have already written?

Give the AI a record of what you've said and check new briefs against it. A memory that keeps each idea, claim and story with its source words can score a brief as a repeat, a build, or something new. Add finished pieces back regularly so that record stays current.

## Methodology

What ScriptGrain keeps is described from the app's code on 2026-10-02 (the Library, Drafts, brief templates, Voice Brain, the voice monitor and team sharing), including what it does not do.

The other kinds of tool are described by what they are built for, in general terms. Products are named as examples of a kind, not reviewed, and nothing here says what any named product lacks.

## Sources

- [ScriptGrain: Voice Brain](https://scriptgrain.com/voice-brain)
- [ScriptGrain: API and MCP server](https://scriptgrain.com/ai-writing-api-and-mcp-server)
- [ScriptGrain: content governance for written content](https://scriptgrain.com/reference/content-governance-for-written-content)

## More reference

- [Content governance for written content: a checklist for voice, claims and repetition](https://scriptgrain.com/reference/content-governance-for-written-content)
- [The measurable markers of AI-generated text](https://scriptgrain.com/reference/markers-of-ai-writing)
- [The words ChatGPT overuses, with the evidence](https://scriptgrain.com/reference/words-chatgpt-overuses)
