Brooks's Law says adding people to a late project makes it later. The mechanism is not the new people, it is the cost of bringing them up to speed while the existing team absorbs the onboarding. The law holds for teams too, and it has a corollary: the faster you onboard the new team, the faster the productivity recovery.
The onboarding documents that matter most are the ones the new team cannot find themselves: the decisions that were made and why, the dependencies that are already committed, the risks that are being tracked, and the things the team tried that did not work. That knowledge lives in the heads of the existing team and the decision log, nowhere else.
I have added teams mid-program twice at significant scale. The pattern that worked both times was a dedicated two-day onboarding that covered scope, decisions, dependencies, and current risks in that order. Not a reading assignment. A facilitated session where the new team could ask questions while the existing team was in the room. Questions answered in that session stay answered. Questions answered in separate emails get answered differently by different people.
Set the new team's first milestone for two weeks out, not immediately. The first two weeks are orientation, even if the team is senior and experienced. Fighting that buys you nothing and costs you a rocky start.
A new team added cleanly is a genuine multiplier. A new team added sloppily costs the program a month before it contributes a week.