Scope is an agent skill by Raphael Salaja that keeps an edit to what you asked for. It reads the smallest literal meaning, keeps what already worked and accounts for every item in your message.
What it does
- Splits your message into numbered asks, adds open items from earlier and reports each one as done or not.
- Changes only the element you named, and asks one short question when a request has two readings.
- Rereads your earlier corrections and keeps values you already gave.
- Lists existing features before a refactor and checks each one after.
- Doesn't push work into a later phase, edits nothing while planning and answers "how does this work" with an explanation.
When to use it
Your agent applies it on every edit. It matters most on short, pointed requests and on messages with several asks.
How to get it
The skill is free on GitHub under the MIT license, in Raphael Salaja's skills repo. Install the set with npx skills add raphaelsalaja/skills.
more skills
Scope is an agent skill by Raphael Salaja that keeps an edit to what you asked for. It reads the smallest literal meaning, keeps what already worked and accounts for every item in your message.
What it does
- Splits your message into numbered asks, adds open items from earlier and reports each one as done or not.
- Changes only the element you named, and asks one short question when a request has two readings.
- Rereads your earlier corrections and keeps values you already gave.
- Lists existing features before a refactor and checks each one after.
- Doesn't push work into a later phase, edits nothing while planning and answers "how does this work" with an explanation.
When to use it
Your agent applies it on every edit. It matters most on short, pointed requests and on messages with several asks.
How to get it
The skill is free on GitHub under the MIT license, in Raphael Salaja's skills repo. Install the set with npx skills add raphaelsalaja/skills.
more skills
Scope is an agent skill by Raphael Salaja that keeps an edit to what you asked for. It reads the smallest literal meaning, keeps what already worked and accounts for every item in your message.
What it does
- Splits your message into numbered asks, adds open items from earlier and reports each one as done or not.
- Changes only the element you named, and asks one short question when a request has two readings.
- Rereads your earlier corrections and keeps values you already gave.
- Lists existing features before a refactor and checks each one after.
- Doesn't push work into a later phase, edits nothing while planning and answers "how does this work" with an explanation.
When to use it
Your agent applies it on every edit. It matters most on short, pointed requests and on messages with several asks.
How to get it
The skill is free on GitHub under the MIT license, in Raphael Salaja's skills repo. Install the set with npx skills add raphaelsalaja/skills.

