This is a fun question! But God is it complicated. If I had to answer it with one word, it would be “SIOP.” Which is technically cheating. Because it’s an acronym.

It's short for SINGLE – INTEGRATED – OPERATIONAL – PLAN

It was in short, the plan that determined where US nukes were aimed for most of the Cold War.

Prior to the creation of SIOP in 1960, targeting of nuclear weapons was piecemeal and handled with single plans that targeted a few dozen Soviet cities at most. These plans had names like TOTALITY, BROILER, and HALFMOON and they were essentially stopgaps to figure out where to aim the nukes in the case of a mostly conventional war that kicked off some time shortly after the Second World War. The US toyed with a number of different ideas for how to prosecute a nuclear war during this period. President Truman unsuccessfully approached the Soviet Union about nuclear disarmament (The Baruch Plan). After the Soviet refusal to participate in the banning of further nuclear weapon development (The US had poisoned the well, by refusing to commit to disarming the weapons they already had), the US began a major arms program to stockpile warheads. By the early 1950s there were more than one thousand warheads in the US arsenal, but questions over how to use them still persisted. There was a limited push to consider a “No Cities” strategy that would target purely industry outside of cities and hard military targets. Other experts argued for a more broad approach that would limit the Soviet ability to fight a conventional war. Others argued for an even broader approach, which would target the Soviet Union and all its allies and partner states, and knock them below the level of functioning opponents in one massive strike. Eventually this final approach won out, leading to the creation of SIOP.

The first SIOP plan was created in 1960, with heavy input from the RAND Corporation as well as White House guidance. Eisenhower was adamant that the plan be finished before he left office. The plan allocated targets to all 3,000+ US nuclear warheads that were deliverable by aircraft or ballistic missile. The strike would hit the Soviet Union, its satellite states and Communist China. The plan would pair each individual warhead to a target, and many targets were “Strategically Re-serviced” (Overkilled) – in case one or more of the warheads was a dud. More than 20 weapons would be used against Moscow alone. Small cities like Baku, Riga or Tallinn were definitely included as targets. Small states like Albania and Mongolia were included as targets. China, even after the Sino-Soviet split, was included as a target. The taiga forests in northern Russia, which could be used to fuel and supply the Soviet war industry, were included as targets. There was precious little than was not a target. One US concern in regards to SIOP during the late Cold War was the increasingly rapid exhaustion of possible targets as US warhead production outstripped them.

The plan was massive and it could not be altered on the fly. It was not flexible. Even if a non-Soviet state like China declared themselves a non-participant, SIOP could not be done in parts. China, as well as all Soviet satellite states would be struck even in a scenario where they were non-combatants. When US Airforce commander Thomas Power was asked by Secretary Robert McNamara why SIOP targeted the entire country of Albania over the placement of a single Soviet radar station, Power responded, "Well, Mr. Secretary, I hope you don't have any friends or relations in Albania, because we are just going to have to wipe it out.” He really said that.

Because the first SIOP was essentially a binary switch which struck all potential targets in all potential US enemies all at once, it did not satisfy the desire among US politicians for flexibility. This would be partially resolved in the next SIOP, SIOP-63. SIOP-63 came in five brand new flavors of escalation, which stacked cumulatively:

1. Soviet nuclear missile sites, bomber airfields, and submarine tenders.

2. Other military sites away from cities, such as air defenses.

3. Military sites near cities.

4. Command-and-control centers.

5. Full-scale "spasm" attack as described in previous SIOP.

SIOP-63 would be amended and updated until 2003. The Nixon administration placed emphasis on adding more delineated options for small scale strikes and specific war-goals. The Reagan administration focused on beefing up deterrence and giving the US more lethal armaments in the case of a full scale spasm-attack. But the core of the plan remained the target list which paired every single US nuclear warhead with a possible target. The US no longer calls it SIOP; it’s now called “CONPLAN-8022” or “CONPLAN-8044.” We don’t really know the name for sure, let alone what it now contains. It’s heavily classified.

So in summary, while US plans started out more specific and piecemeal, over time they developed into a monstrous contingency plan that targeted the breadth of the Soviet Union, striking population centers, far flung cities, industrial zones, nuclear weapons sites, military bases, small member states and even potential non-combatants. Everything you named in your question would be hit, plus so much more. While more options were added over time for more targeted strikes, the core of the plan continued to be based around a single massive spasm attack even until after the collapse of the Soviet Union. That could still be the core of the plan, we just don't know for sure. As I said at the beginning, it’s a fun question.

Sources:

The Creation of SIOP-62: More Evidence on the Origins of Overkill (gwu.edu)

U.S. Nuclear War Plan Option Sought Destruction of China and Soviet Union as "Viable" Societies | H-Diplo | H-Net (h-net.org)

The actual target lists have never been made public. However, we do have some indications as to what a war-plan from the late 1950s would have looked like. Here are the mapped targets named in a 1956 planning document (more details here). You can see that if such a list were turned into an actual war plan it would be extensive and crippling (an actual war plan would tell us the megatonnage, altitude setting, and number of weapons assigned to each target chosen, as well — it was not "one bomb, one target" for important targets).

Ostensibly it is entirely "strategic" — it is meant to destroy the USSR's ability to wage war. But many of those targets are in urban areas. The USA estimated that 500 million people would be killed from the fallout alone in such an attack. (Daniel Ellsberg's The Doomsday Machine describes these estimates well.)

Note that in 1956, the Soviet Union would have had essentially no serious capability to destroy the USA in turn. It could have wrecked considerable havoc on Western Europe and other American allies. But its ability to reliable deliver nuclear weapons to the continental USA was very limited. Which is to say, it was not mutually assured destruction at this point — that was not until the mid-1960s, at the earliest. Even then, the USA and USSR were not at parity in their capabilities until the 1970s-1980s.