Define a Convex schema with typed field validators and indexes in convex/schema.ts using defineSchema and defineTable

Documented steps

1. Ensure `npx convex dev` is running (it reads CONVEX_DEPLOYMENT and pushes code to your dev deployment) before editing schema files.
2. Create convex/schema.ts and import { defineSchema, defineTable } from "convex/server" and { v } from "convex/values".
3. Export default defineSchema({ ... }) with one object key per table name.
4. Define each table as tableName: defineTable({ field: v.string(), other: v.number(), ref: v.id("otherTable") }) using validators like v.string(), v.number(), v.boolean(), v.optional(v.string()), v.union(...), v.literal(...), v.array(...), v.object({...}), v.record(...), v.any().
5. Add indexes by chaining .index("by_token", ["tokenIdentifier"]) on the defineTable(...) result; for compound indexes list fields in the order they will be queried, e.g. .index("by_channel_user", ["channel", "user"]).
6. Do not manually declare fields starting with "_" (e.g. _id, _creationTime) — Convex adds and manages these automatically.
7. For a circular v.id() reference between two tables, make one side nullable: v.union(v.id("table"), v.null()).
8. Run npx convex dev or npx convex deploy to push the schema; on first push Convex validates every existing document against the new schema and halts the deploy if any document does not match.
9. To bypass runtime document validation temporarily, pass { schemaValidation: false } as the second argument to defineSchema.
10. After a successful push, import the generated Doc<"tableName"> and Id<"tableName"> types from convex/_generated/dataModel for typed access in your functions.
11. Official docs: https://docs.convex.dev/database/schemas and https://docs.convex.dev/database/reading-data/indexes/