IMO your choice of stack is fairly subjective and also depends on what types of problems you're trying to solve in each projects. Every library will have pros and cons that you'll have to weight against your projects requirements.
Now that being said, here's a few library that I like to keep in my tool belt .
- Autofac - IoC Container. Very well documented, great platform support, and tons of 3rd party integrations
- FluentValidation - Object validation library. Works well in web projects. Lets you define reusable model validation rules.
- xUnit - Unit testing library. Very extensible and lightweight
- FakeItEasy - Mocking library. Uses a fluent syntax.
- Dapper - Not an ORM but does map objects to SQL tables. Add extension methods to the IDbConnection interface. Very good performance.
- Entity Framework 6 - Solid and Mature ORM from Microsoft.
- RethinkDB Drive for .NET - .NET support for realtime database Rethinkdb. Works across almost every flavor of .NET
- Bogus - Fake Data generator. Good for testing / prototyping.
- ASP.NET Core - Cross platform/Open source version of ASP.NET. Lightweight, fast, and more editor friendly.
- JSON .NET - Popular JSON Serializer.
- Serilog - Library for structured logging.
Again these are the ones that I like to use in both large and small project. Your preferences may vary as well as the requirements of your project. What you really should focus on is the architecture of your solution, .i.e what's the best way to create a performance, modular and maintainable product. Libraries of come and go from time to time.