Akita Finish logo

Akita Finish

Community
akitaonrails
akita-finish

Final readability and voice pass for Akita's blog articles (PT-BR canonical, EN sibling). Run AFTER the humanizer skill on every new article draft and on any substantial edit, before showing the text to Akita. Enforces short paragraphs, lists for enumerations, blockquotes for takeaways, ### subsections, italics/bold conventions, rounded numbers in prose, and the voice bans (no "não é X, é Y" tails, no honesty/directness signaling, no process asides, no effusive flourishes).

Overview

Publisherakitaonrails
Repositorymy-skills
Skill nameakita-finish
Stars
276
Forks
30
Bundled files
Instructions only
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.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

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

Installation

Install the Akita Finish 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/akitaonrails/my-skills/tree/master/akita-finish
  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/akitaonrails/my-skills.git /tmp/my-skills
mkdir -p .claude/skills
cp -r /tmp/my-skills/akita-finish .claude/skills/akita-finish
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Akita Finish 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 Akita Finish 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 Akita Finish 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.

akita-finish: finishing touches for an article

This is the third pass of the article pipeline, applied to the file on disk:

  1. Draft the article from the brief and the sources.
  2. fact-check: adversarial verification of every claim, number and attribution against the sources. Run the fact-check skill's checker prompt (references/checker-prompt.md) inside isolated general-purpose agents in this session: one agent for claims verifiable against local source files, one or more for web claims. Do not use the external CLI harnesses (grok/zcode/codex) unless Akita asks. Apply the corrections.
  3. humanizer: remove AI-writing tells.
  4. akita-finish (this skill): structure, rhythm, emphasis, numbers and voice bans.

Run it as an explicit pass over the finished text. Do not rely on having "kept the rules in mind" while composing; count and grep.

Pass 1: structure

  • Paragraph ceiling: 4 sentences. Count sentences in every prose paragraph (skip tables, lists, blockquotes, headings). Split anything with 5 or more at the beat boundary. Short paragraph after long paragraph is part of the voice.
  • List-shaped prose becomes a list. A paragraph that enumerates 4+ parallel items (use cases, tasks, conditions, metrics, options, steps) becomes a bulleted or numbered list. Keep one short intro sentence before it.
  • Blockquotes for takeaways. Put the key conclusion of a section, the TL;DR, warnings and "what this means for you" in a > blockquote. Also use one to break any stretch of 6 to 8 consecutive plain paragraphs with no visual break.
  • ### subsections to scope any ## section longer than roughly 6 to 8 paragraphs. Headings in sentence case, no Title Case.
  • Narrative flow per section: the point first, then the evidence, then what it means for the reader.

Pass 2: emphasis and numbers

  • Italics for titles of cited works (books, essays, papers, articles) and for short quoted phrases used as names.
  • Bold for important names, products and terms on first mention, and for the single key number in a table. Never bold a whole sentence, never bold-label every list item.
  • Round numbers in prose. "quase 100" beats "99,876"; "mais de 200" beats "211,59"; "uns 14%" beats "13,8%". Keep the exact figure only inside tables, spec sheets and code blocks, or when the exact digit is the point (a price list, a measured record). Prose gives the order of magnitude; tables give the precision.
  • Ratios over raw pairs. "o dobro da velocidade" beats "A levou 10 ms e B levou 20 ms"; "três vezes o preço" beats two dollar figures side by side; "um erro a cada 40" beats "97,6%". Give the reader the picture; leave the two raw numbers to the table.
  • Numbers in PT-BR use decimal comma and thousands point (1.349 ms, R$ 5,22).
  • Keep every inline source link. Add one when adding a claim.

Pass 3: voice bans (grep, then read)

Run these greps over the file and fix every hit:

sh
grep -n "—\|–\|“\|”\|‘\|’" FILE                      # dashes and curly quotes: none allowed
grep -n -i "não é [^.]*, é \|, e não \|, não [a-z]*\.$\|, nunca " FILE   # negative tails
grep -n -i "pra ser honest\|sincer\|franco\|justiça seja\|vou ser direto\|preciso ser direto\|sem rodeios\|cirúrgic\|verdade seja dita\|devo admitir" FILE
grep -n -i "honest[ao]\|honestidade" FILE            # "honest" as emphasis ornament: banned
grep -n -i "refetchável\|deletável\|fetchável" FILE  # invented anglicism calques: banned
grep -n -i "assino embaixo\|suculent\|cutuc\|lindo de" FILE   # effusive flourishes

Then read for what grep cannot catch:

  • "não é X, é Y" and "X, e não Y" written with other words. Rewrite as the affirmative alone.
  • Process asides ("X insistiu nisso, e concordo", "o modelo fez questão de"). State the finding.
  • Announcing honesty or directness in ANY form ("em lista honesta", "a honestidade primeiro", "seja franco"). The voice is straightforward by default, so every "honesto" is filler. Delete it.
  • Invented anglicisms: English words dressed as Portuguese ("refetchável", "deletável", "escapável" for "escapar"). If the word is not natural PT, rephrase plainly ("que dá pra baixar de novo").
  • Rule-of-three padding, false ranges, "serve como/representa" instead of "é".
  • The same anglicism three or more times in nearby paragraphs; translate it.
  • Refer to the author as "Akita" in third person where needed, never "Fabio".

Pass 4: report

Tell Akita in two or three lines what the pass changed (how many paragraphs split, lists created, blockquotes added, numbers rounded, voice fixes). Then stop for review. The next step (translation, index regeneration, commit) only happens after he approves.

Frequently asked questions

What does the Akita Finish AI skill do?

Final readability and voice pass for Akita's blog articles (PT-BR canonical, EN sibling). Run AFTER the humanizer skill on every new article draft and on any substantial edit, before showing the text to Akita. Enforces short paragraphs, lists for enumerations, blockquotes for takeaways, ### subsections, italics/bold conventions, rounded numbers in prose, and the voice bans (no "não é X, é Y" tails, no honesty/directness signaling, no process asides, no effusive flourishes).

Why use Akita Finish on TypingMind?

Because you install it once and use it with any model. Akita Finish 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 Akita Finish in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/akitaonrails/my-skills/tree/master/akita-finish. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Akita Finish?

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 Akita Finish?

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

Is the Akita Finish AI skill free?

Yes. It is published on GitHub by akitaonrails 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 👇