This will be done, if ever it is needed, by depositing hundreds of millions of tons of reflective material (basically refined moon dust) into Lagrange Points which will block a proportionate amount of energy from reaching the Earth.
The solar energy output from the sun (measured as 'total solar irradiance' or TSI) is not constant. It ebbs and wanes in a fascinating quasi-periodic fashion on daily, 11-year, 210-year, 350-year, etc. cadences. It also shows an overall increasing trend over millennia timeframes (3 - 4 billion years ago the sun output only ~70% of its current energy). Since about the 1940s up until quite recently we have been experiencing a period considered a "modern grand maximum" in solar activity. [1, 2]
All this is simply to show that the sun is not an absolute constant, as Aristotle believed, but a dynamic and fluctuating system which directly impacts -- actually that understates it, variance of solar output has been the primary driver of our wildly varying global climate for the Earth's entire history. Which makes it all the more notable that anthropogenic climate change has now begun to bend the curve in similar fashion to our mighty star.
Crucially, the technological moderation of TSI is not a risky endeavor even remotely like detonating nuclear bombs to dust up the atmosphere. This would not be a last-ditch gambit. The impact of adding reflective material between the Earth and sun is miniscule on the scale of how much material could be placed into position at a given time, e.g. using rail-guns on the surface of the moon. It is a multi-decade effort that requires constant upkeep to maintain, or otherwise naturally dissipates over time. It is easy to measure precisely the impact achieved and therefore precisely control how much of a hand we put on the celestial scale. And once established and in place, e.g. 50 years from now, will cost likely on the order of $500 billion per year in upkeep. That is to say, an absolute bargain in terms of the economic impact of doing nothing.
It is neither physically impossible, nor constrained by the practical extraction of natural resources that would need to be brought to bear, nor dangerous, nor requires exotic undiscovered materials or magical technology that could not be fabricated given proper funding over the next couple decades.
I have no doubt or concern that our technological capability for generating clean energy will continue to grow exponentially, hopefully culminating in abundant and asymptotically free, clean energy within the century. And I have no doubt that while reaching that crowning achievement that humanity is entirely capable of either extracting excess CO2 directly from the atmosphere, or moderating TSI in order to achieve an optimal climate equilibrium.
This is by no means easy, or even guaranteed. But I firmly believe that we are not doomed, and that we remain entirely in control of our fate. In my opinion, the solution will not be found in the direction of policies that tax, ration, or restrict abundant energy, for that is an unethical chokehold that will be felt most directly by the poorest peoples among us.
We must strive for technological solutions that make energy cleaner, cheaper, reliable, safe, and abundant for energy is truly the path out of poverty. We have fewer humans living in poverty than at any time in human history [3], and we must continue on that path.
I think as our energy and space technology continue to rapidly advance, the hardest part will be deciding not how to set the thermostat, but exactly where to set the thermostat, because there will be winners and losers regardless.
[1] - https://link.springer.com/article/10.1007/s41116-017-0006-9 [2] - https://faculty.wcas.northwestern.edu/~infocom/The%20Website... [3] - https://ourworldindata.org/extreme-poverty