#Saved skills
Teach Arachne a way of doing something once, give it a name, and call it back whenever you need it.
#Overview
A saved skill is a way of doing something that you have given a name to, so you can invoke it later without re-explaining it. When a piece of work is something you will reach for again — a particular kind of research, a build you keep repeating, a flow you walk through often — you capture how it is done as a skill and call it back by name whenever you want it.
This matters because good work is rarely a one-off. Once you have described an outcome clearly and watched Arachne reach it the way you wanted, that knowledge should not have to be rebuilt from scratch the next time. A saved skill keeps it ready, so routine work happens on demand instead of starting from zero each time you need it.
#What you can do
- Save a way of doing something as a reusable skill with a name you choose.
- Invoke a saved skill later by that name, instead of describing the whole task again.
- Reuse the same skill across new inputs — the same approach applied to a fresh subject each time.
- Keep a small library of the things you do often, so they are there when you need them.
- Combine a skill with the permission levels, so an invoked skill runs at the level of freedom you decide for it.
- Refine a skill in plain language when what you want changes, and keep using it under the same name.
#Skills versus automations
Skills and automations both spare you from re-explaining work, but they answer different needs.
| Saved skill | Automation | |
|---|---|---|
| What it is | A named way of doing something you invoke yourself | A repeatable task Arachne can carry out for you |
| How it runs | You call it by name when you want it | It runs on demand as a task you have set up |
| Best for | A reusable method you apply to new inputs | Routine work you want to repeat without re-explaining |
| You provide | The name, plus whatever the task needs this time | The task itself, ready to run again |
The simplest way to hold the difference: a skill is a method you keep and invoke by name; an automation is a task you keep and run again. Many people use both — a skill for the way they like a thing done, an automation for the things they want to happen on a regular cadence. For repeatable tasks, see Automations.
#How to use it
- Do the task first the ordinary way: describe the outcome you want to Arachne in plain language, watch it work, and steer until the result is right.
- When you are happy with how it was done, save that way of doing it as a skill and give it a clear, memorable name — something you will recognize later, like the kind of work it does.
- Later, invoke the skill by name. Supply whatever is specific to this run — the new subject, link, or constraint — and let Arachne apply the saved method to it.
- Choose a permission level for the run that fits the task, the same way you would for any Arachne work: cautious when the inputs are new, freer when the skill is well understood. See Permissions & control.
- Watch the first invocation so you can confirm the saved method still does what you intend, and steer if anything has changed.
- If your needs shift, refine the skill in plain language and keep invoking it under the same name.
Name skills for the outcome, not the steps — a name like "weekly competitor scan" tells you what you get, while a name about how it works ages badly when the method changes.
#Examples
What you ask
Save the way you just did that competitor pricing scan as a skill called "pricing scan."What you see: Arachne confirms the method is saved under the name pricing scan, ready to invoke later. The next time you ask for it, you only name the skill and the competitor you want scanned.
What you ask
Run my "pricing scan" skill for this new competitor.What you see: Arachne applies the saved approach to the new subject and returns a comparison — your options laid out side by side against the things you care about — without you re-describing how the scan should be done.
What you ask
Use my "weekly news roundup" skill, but only cover the last three days this time.What you see: Arachne runs the saved method with your one change applied, and delivers the roundup as a report with the references gathered into a source pack so you can check any claim.
#Tips
- Save a skill only after a run came out the way you wanted — you are capturing a method that works, not a first attempt.
- Keep skill names short and descriptive so you remember them and can invoke them quickly later.
- Tell Arachne what changes for this run when you invoke a skill — the new subject, link, or limit — and let the saved method handle the rest.
- Watch the first run of a freshly saved skill, and watch again after you refine one, so you can confirm it still does exactly what you intend.