I have worked on identity systems for a good part of my career and I still lose track of what someone should read next. The specs live in working groups that rarely reference each other, and none of them tell you where they sit in relation to the rest.
Then agents arrived and the field grew again. An identity engineer now has to reason about software that acts on someone’s behalf, often across several apps at once. When a new teammate asked me where to start, I had no single place to send them, so I built one. It lives at github.com/alvaroinckot/identity-engineering-roadmap.
I borrowed the roadmap.sh format. You walk a spine from the first primitive to the daily practice of reading specs, and you take the side branches when your job needs them. Today it holds 14 areas and 85 topics, and I keep adding.
Each topic fits on one page. I open with a short paragraph for someone new to identity, then I write the dense version, including what changes when the caller is an agent. Resources close the page, tagged so you can tell the spec from the explainer. I read the spec first and I suggest you do the same.
I did not allow myself to write from memory. If I could not find a source to cite for a claim, the page says so and waits for one. Writing this way took much longer than I expected, and it is the reason I trust the map enough to put it in front of other people.
If you spot a gap or a wrong resource, open a pull request. I want this to outgrow my own reading list.