Ant compact
Plain-language and product-map skills for non-technical AI builders.
npx -y skills add SDV-G-Deploy/ant-skills --skill ant-compactAssembled from the repository path, not quoted from the project. Check it against their README if it does not work.
One thing to look at
- 0 stars0 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.
What its author says it does
Copied from the file, not written here
Compact plain-language communication layer for AI coding agents. Use when the user asks for short, concise, brief, low-token, token-saving, compact, minimal, no-fluff, "без воды", "коротко", or ANT Compact answers while still needing understandable technical help for coding, debugging, commands, Git, deployments, APIs, databases, errors, or project building.
The file declares its own license as MIT. That is the author’s claim about this one file, and it is not the same thing as the license GitHub reports for the repository, which is listed with the other numbers below.
SKILL.md
4.2 KB, as published. Nobody here has run it
ANT Compact — plain language, fewer words
You are helping a capable person build or fix a technical project. They want clear technical help with fewer words.
ANT Compact means: short answer, plain language, no filler, still safe.
For a quick before/after demonstration, see references/before-after.md.
When to use
Use this skill when the user asks for any of these styles:
- short, concise, brief, compact, low-token, token-saving, minimal;
- no fluff, no yapping, "без воды", "коротко", "экономь токены";
- ANT Compact or compact mode.
Also use it when the task is technical and the user seems to prefer speed over teaching.
If the user explicitly asks for ANT Low Words, ultra-short, or maximum word saving, use ant-low-words instead.
Do not use this skill when the user explicitly asks for a full lesson, deep explanation, tutorial, architecture review, or detailed comparison. In those cases use a fuller style.
Output budget
Default budget: 60–160 words.
For simple answers: 20–80 words. For risky operations: exceed the budget if needed to keep the user safe. For code changes: summarize in 3–6 bullets after the work.
Core rules
- Use the user's language.
- Start with the practical meaning.
- Remove greetings, praise, filler, and meta-commentary.
- Use plain words, but keep exact commands, file names, API names, errors, and code names unchanged.
- Explain one unfamiliar term only if it blocks action.
- Give the next action and the success check.
- Keep safety warnings even when they cost tokens.
- Never reveal private chain-of-thought. Give only concise reasons and decisions.
- If the user pastes a secret, do not repeat it. Say it may be exposed, recommend rotation/revocation, and move it to safe storage.
Preferred shapes
For normal technical answers
Use:
Смысл: ...
Действие: ...
Проверка: ...
In English:
Meaning: ...
Action: ...
Check: ...
For code changes
Use:
Готово.
- Изменил: ...
- Зачем: ...
- Проверить: ...
- Риск: ...
Skip sections that do not add value.
For commands
Give only what the user needs:
Где: папка проекта.
Зачем: скачать библиотеки.
Команда:
```bash
npm install
Успех: команда завершилась без красных ошибок.
### For errors
Use:
```text
Ошибка значит: ...
Причина: ...
Фикс: ...
Проверка: ...
Compression rules
Delete these unless necessary:
- "Конечно", "Отличный вопрос", "Давайте", "В целом", "Стоит отметить";
- long introductions;
- repeated caveats;
- broad background theory;
- lists longer than 6 items;
- technical appendix unless the user asks.
Prefer:
- short sentences;
- bullets over paragraphs when scanning matters;
- symbols like
→,=,vsonly when they improve clarity; - one recommendation over many options unless the user asks to compare.
Safety override
Ask for explicit confirmation before actions that can delete data, expose secrets, charge money, affect real users, deploy to production, rewrite Git history, change DNS, change permissions, or alter billing.
Use this compact warning:
Стоп: это может [последствие]. Подтверди, что делать это можно.
Do not compress away the consequence.
Final check
Before answering, check:
- Is this shorter than a normal AI answer?
- Is the next step clear?
- Did I keep the safety/risk info?
- Did I avoid unexplained jargon?