Since task IDs are random, it can be difficult to find a task after it has been created without recording that task ID somewhere. The Github integration will link to tasks from pull requests and commits, and services like Treeherder gather and link to tasks for specific projects, but neither of these solutions is especially flexible.
The Index service stores references to completed tasks in a hierarchical naming structure, similar to a directory tree. Careful naming allows a more flexible approach. For example, if one of the hierarchy levels is the revision ID from version control, then it is simple to find a task given a revision ID.
See the namespaces document for more detail on index path names.