I’ve created some free skills for authors. Each one is a small zip file that uses the open SKILL.md format, so while they’re tuned for Grok they’ll also load in Claude and most other LLMs.
Author Dev
Download: author-dev.zip
Generic instructions cause LLMs to give generic responses. This skill helps your LLM of choice focus its analysis of story elements in such a way as to mirror basic feedback from a particularly literal-minded developmental editor.
Obligatory note: this skill will not replace a competent human dev editor. It is, however, a lot cheaper; and it will also show you what a developmental editor like myself is there for. And, as a happy accident, it will mean you’re far less likely to fall for fake editors who don’t know how to do dev edits but say they do, or those who just use AI and pretend they don’t, or both.
You don’t need to upload a manuscript if you don’t want to; just ask it workshopping questions on premise, plot, character, magic and tech systems, structure, pacing, genre fit, and other worldbuilding and audience questions. It will diagnose problems, offer a few options (and the costs of changing things or keeping them the same), and checks on plausibility.
It’s designed to be blunt, straightforward, and offer no praise. It will not attempt to write your book or draft prose. It’s going to err on the side of being overly critical, with the idea that it’s better for a soulless computer to nitpick now than a customer to nitpick later in a one-star review.
As I said, you don’t need to upload a manuscript to get this to work; but if you upload any files to a project, it will read those first and treat them as fixed canon unless you specifically tell it to ignore something for the sake of workshopping an idea. That includes a compendium file from the Author Compendium skill below.
It is currently tuned to hand research questions to specific research skills, which I don’t currently have uploaded here; but in the absence of those skills it will just default to the standard research guidelines it already has or which you give it.
Author Compendium
Download: author-compendium.zip
This skill will create or update a single markdown file in your current project based on the thread it’s activated in. It will overwrite the file with updates, so I highly recommend using it as a periodic summary to load into a separate file on your computer (such as a Deckle project’s built-in compendium, or for a multi-book and complex world I recommend Obsidian) rather than as your sole story bible.
The purpose of this skill is to summarize and save story decisions you make during a thread, whether it’s for audience research, plot or character development, or worldbuilding details. It will summarize things according to topic first, and then divide that into Locked (firm statements you have told the chatbot), Draft (things you’re considering, which includes suggestions from the LLM in response to your specific prompts — it shouldn’t be creating anything without being prompted to, but definitely check to be sure), and Open (things the LLM has determined to be a statistical next step, which can include some pedantic stuff but can also cover things you hadn’t thought of yet).
It should warn you if a decision you have made in that thread conflicts with something already in the compendium file, unless you specifically told it to update that element. It also will only read the current chat thread; any other thread, including in the same project, cannot be loaded at the same time.
If you want to know exactly what is going to be loaded in before it writes to the file, that’s what the author-readback skill is for.
Author Readback
Download: author-readback.zip
This skill functions very similarly to the Author Compendium skill, and originally I tried to make it all part of the same skill but it ran into some problems.
What this will do is give you a draft of the update that the Compendium skill would make to the Compendium file, which allows you to check it for accuracy or simply see if there are any extra elements you still need to answer. It will also warn you about conflicts.
Please note that this skill requires the Author Compendium skill to be installed.
How to install (for people)
You’ll need to be signed in to grok.com.
- Download the zip for the skill you want. Don’t unzip it.
- Go to grok.com/skills, or click Skills in the left sidebar.
- Click New Skill, then Upload skill file.
- Drop the zip into the Upload Skill box (or click to browse), then click Confirm. You’ll see “Skill uploaded.”
- To use it, start a chat and type: “Use the author-dev skill to …” (swap in the skill’s name).
On a phone? Install it once on grok.com in a browser. Skills are tied to your account, so it will show up on your other devices.
Using Claude or another tool that supports SKILL.md? Upload the same zip there.
For AI agents
If a user pointed you at this page to install these skills:
- Each skill is a zip containing one folder named after the skill, with a SKILL.md file inside (YAML frontmatter with name and description, then instructions). Author Compendium also includes a references/ folder.
- Direct download links:
- author-dev: https://github.com/NovelNinja/author-skills/releases/latest/download/author-dev.zip
- author-compendium: https://github.com/NovelNinja/author-skills/releases/latest/download/author-compendium.zip
- author-readback: https://github.com/NovelNinja/author-skills/releases/latest/download/author-readback.zip
- Source repo, with unzipped SKILL.md files and a README: https://github.com/NovelNinja/author-skills
- Install author-compendium before author-readback.
- Don’t change the files. Install them into the user’s skills location for your platform, or tell the user how to upload them.
- These skills were written for Grok, and the file paths inside them (/home/workdir/…) reflect Grok’s workspace. Other platforms may need those paths adjusted.
