Software shouldn't decide what you can do with it. Say which program you run and what you want from it — we point you to a skill for that app, your agent does the work on your machine, and you check the result.
Independent guides. Coverage is limited; each guide page shows what has been checked.
Start with the app you use and the result you want. Check whether a guide is available before you download.
Start with the software you already use. Each guide explains its scope, what the downloaded skill gives your agent, and what to verify before you rely on a result.
Some behavior is a setting. Some needs a different route. Some can't be changed at all — and knowing that early saves hours. Start with what you want, not with how the code works.
Ask about something specific you can describe: why it slows down, why it changed, what controls it. You get an explanation instead of a guess.
Some behavior can be adjusted and some can't, and knowing which one you're looking at is the useful part. You decide what to do with that answer.
When a feature disappears, the honest answer may be a setting, another route, or a different tool that does the job today. Finding out which is the point.
Tell us the software and your goal. Get the guide, hand your agent the starting request, and check the result against the evidence.
Enter the software you run. A real match takes you to its guide page; an uncovered app gets an honest answer.
Download the guide, follow its file-handoff steps and add your own goal. Setup depends on your agent and task.
Ask for the evidence behind the result. Keep the useful findings, and see what remains unverified.
Not one fixed feature. It gives your agent context for that app — where things live, what breaks, how to verify a claim — so it works on the goal you typed, not a demo.
The right starting material and a route into the target you actually use.
Bring the change, question or reconstruction you want to explore.
Keep the evidence and the limits visible as the work progresses.
We don't take uploads, don't change code for you, and don't cover every app. If no guide exists for yours, we say so. Your agent's model is a separate service with its own data rules.
Direct answers first. If something here doesn't apply to your case, that's information too.
It depends on the app and your setup. Each guide page states what has been checked, its target and any unverified steps. A download does not guarantee a successful analysis of your version.
Enter the software and what you want to do. If a guide exists, open its page to download the package and follow the handoff steps. If it doesn't, we tell you.
We show that no guide is available and link to the official REA setup. There is no placeholder download or notification sign-up pretending to be ready.
You describe the goal in plain words and the skill gives your agent the context for that app. Setup depends on your agent and task; some tasks still call for technical judgement.
We don't accept uploads. Your agent inspects the software in your environment. Its model service may receive information the agent reads or sends; check that provider's data rules.
No universal success is promised. Availability means the guide package exists, not that every target or task has been tested. Read its coverage and verification notes before you start.
Tell us what you run and what you want from it. Find a real guide, bring your goal, and take the next step with your agent.