Non Autoregressive Writing Pass logo

Non Autoregressive Writing Pass

Community
msimchowitz
non-autoregressive-writing-pass

Revisit a completed draft with global knowledge of its argument, then revise its title, subtitle, headings, paragraph heads, and paragraph-to-paragraph transitions. Use when display text no longer fits the finished work; when openings feel generic, abrupt, weak, or begin with a rejected alternative before reaching the point; when paragraphs are individually clear but do not build on one another; when a finished abstract needs a stronger first sentence; or when final sentences do not close one paragraph naturally or prepare the relation to the next.

Overview

Publishermsimchowitz
Repositorywriting-skills
Skill namenon-autoregressive-writing-pass
Stars
67
Forks
4
Bundled files
1
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • 1 bundled files

    Scripts, templates, and references the model can read while it works. Files are read-only and never executed.

  • Open source

    Published by msimchowitz on GitHub. Read the source before you install it.

Installation

Install the Non Autoregressive Writing Pass AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/msimchowitz/writing-skills/tree/main/for-agents/non-autoregressive-writing-pass
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/msimchowitz/writing-skills.git /tmp/writing-skills
mkdir -p .claude/skills
cp -r /tmp/writing-skills/for-agents/non-autoregressive-writing-pass .claude/skills/non-autoregressive-writing-pass
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Non Autoregressive Writing Pass in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Non Autoregressive Writing Pass on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Non Autoregressive Writing Pass is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Non-Autoregressive Writing Pass

Revise the completed draft with access to the whole argument. In this skill, "non-autoregressive" describes the editing pass: do not accept a sentence merely because it follows smoothly from the text before it. Judge each opening and transition against the material that comes after it as well.

Wait Until The Draft Is Stable

Run this pass after the claims, evidence, paragraph order, and broad cadence are settled. It is not a drafting outline or a substitute for structural revision.

Read the complete artifact first. Identify its central claim, the role of each section, and the point each paragraph eventually establishes. Preserve facts, citations, qualifications, notation, examples, and deliberate voice.

Read The Title And Paragraph-Head Spine

Extract the title, subtitle, every section and subsection heading, the abstract opening, and the first one or two sentences of every paragraph. Keep the headings in place and read this sequence without the paragraph bodies. Treat the first two sentences as one paragraph head when the first supplies context that the second turns into a claim.

For the title, subtitle, and headings, ask:

  • Does the title name the subject or tension that actually organizes the finished work?
  • Does the subtitle sharpen that account instead of repeating it or preserving an earlier emphasis?
  • Does each heading accurately frame the material beneath it without sounding like a slogan, instruction, interface label, or generic category?

For each abstract or paragraph head, ask:

  • Does it state or naturally approach the point that the paragraph actually develops?
  • Does it rely on a vague pointer such as "This," "The same," or "These issues" when the referent should be named?
  • Does it repeat the heading, announce the document, or begin with a generic truth that could introduce many subjects?
  • Does it begin by rejecting an alternative before stating the substantive claim?
  • Does it inherit the previous paragraph's syntax or vocabulary merely because that wording was locally available?
  • Does the abstract begin with the concrete problem or tension that organizes the work, rather than a weak generality?

A topic sentence need not contain the paragraph's whole conclusion. It may bridge from the previous paragraph, establish necessary context, or create a natural opening before the main claim arrives in the second sentence. Revise the first two sentences together when forcing everything into the first would make the prose stiff.

When the paragraph's main claim is positive, state it before explaining what it is not. An opening such as "The difference is not X. The difference appears in Y" makes the reader retain a rejected frame before reaching the point. State Y first, then mention X only if the contrast still matters. Keep a negative opening when an absence, failure, or limitation is itself the finding or claim boundary.

Inspect Every Paragraph Boundary

First place each title, subtitle, or heading beside the opening it frames. Then place the final sentence of each paragraph beside the first two sentences of the next. Name their relation: continuation, cause, consequence, contrast, qualification, example, change of scale, or change of question.

Revise a boundary when:

  • the title, subtitle, or heading promises a subject that the opening does not develop;
  • the ending merely recaps the paragraph or appends a transition whose real purpose belongs in the next opening;
  • the next paragraph restarts the subject as though no prior reasoning occurred;
  • a pronoun or generic transition hides the relation;
  • both sentences repeat the same conclusion;
  • the handoff implies a stronger causal or evidential relation than the draft supports; or
  • a section break interrupts a thought that should close before the heading.

Revise the ending, the next opening, or both. Prefer lexical continuity and a clear logical relation to formulaic connectors such as "Additionally," "However," or "With this in mind." Do not add a transition when adjacency already makes the relation clear.

Let Endings Finish Their Own Reasoning

A paragraph's last sentence should first complete that paragraph. It may end on a consequence, boundary, unresolved tension, or concrete fact. It can also prepare the next paragraph, but it should not become a trailer for the document.

When the next paragraph changes direction, let its opening carry the turn. When the next paragraph elaborates the same idea, a repeated term may provide enough continuity. Do not manufacture a smooth handoff by weakening a meaningful break.

Run The Final Read

  1. Read the revised title-and-paragraph-head spine again. Confirm that its titles, headings, and openings trace the argument without sounding like an outline.
  2. Read every ending-to-opening pair at speaking pace.
  3. Read the full artifact in order and restore any deliberate abruptness or delayed reveal that the isolated checks flattened.
  4. Compare with the source and confirm that the pass changed no claim, uncertainty, evidence, or scope.

Invocation Boundary

When another writing skill invokes this pass, return only the revised prose or the requested boundary findings. Do not invoke Humanizer, general-writing, writing-cadence, better-usage, or academic-voice; the calling skill owns the larger prose pipeline.

Bundled files

The model reads these on demand while the skill is loaded. They are exposed as readable files and are never executed.

Frequently asked questions

What does the Non Autoregressive Writing Pass AI skill do?

Revisit a completed draft with global knowledge of its argument, then revise its title, subtitle, headings, paragraph heads, and paragraph-to-paragraph transitions. Use when display text no longer fits the finished work; when openings feel generic, abrupt, weak, or begin with a rejected alternative before reaching the point; when paragraphs are individually clear but do not build on one another; when a finished abstract needs a stronger first sentence; or when final sentences do not close one paragraph naturally or prepare the relation to the next.

Why use Non Autoregressive Writing Pass on TypingMind?

Because you install it once and use it with any model. Non Autoregressive Writing Pass is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Non Autoregressive Writing Pass in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/msimchowitz/writing-skills/tree/main/for-agents/non-autoregressive-writing-pass. TypingMind reads its SKILL.md and bundles its files and installs it as a skill you can enable per chat.

Which AI models can use Non Autoregressive Writing Pass?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Non Autoregressive Writing Pass?

As many as you like. As long as a model supports skills, you can use Non Autoregressive Writing Pass with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Non Autoregressive Writing Pass AI skill free?

Yes. It is published on GitHub by msimchowitz under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇