Validate Convex function arguments and return values using the v validator builder's args and returns fields

Documented steps

1. Import { v } from "convex/values" and query/mutation/action from "./_generated/server".
2. Pass an args object of validators to the function definition, e.g. args: { body: v.string(), author: v.string() }.
3. Optionally add a returns validator describing the output, e.g. returns: v.null() for a mutation with no meaningful return value, or returns: v.id("tasks").
4. Use primitive validators for scalar fields: v.string(), v.number(), v.boolean(), v.null(), v.int64(), v.bytes().
5. Use v.id("tableName") for document ID arguments, v.optional(validator) for fields that may be omitted, v.array(validator) and v.object({...}) for structured data, v.union(a, b, ...) for multiple allowed shapes, v.literal(value) for an exact constant, v.record(keyType, valueType) for dynamic-key maps, and v.any() to opt out of validation.
6. Read the validated, typed arguments from the handler's second parameter; TypeScript types are inferred automatically from the validators.
7. Reuse and compose object validators with their .pick(), .omit(), .extend() and .partial() methods instead of redefining overlapping shapes.
8. Use Infer<typeof myValidator> to extract a standalone TypeScript type from a validator.
9. Apply the same args/returns validation pattern uniformly across public queries, mutations, actions and internal functions.
10. Official docs: https://docs.convex.dev/functions/validation