Roadmap

In no particular order, here are some things We want to implement:

High Priority

• ✅ More assertions
  • ✅ Better / more async support
  • ✅ is true / is not true
  • ✅ is false / is not false
  • ✅ is NaN
  • ✅ Support for more intrinsics; Set, Map, WeakSet, WeakMap, WeakRef
  • ✅ Random convenience like is frozen, is sealed, is extensible
  • ✅ Deep equality / partial equality, strict and loose
  • Callbacks & async callbacks
  • Whatever else. There's always room for more, within reason!
• Custom assertion improvements
  • Custom diffs
  • ✅ Custom error messages (via Zod)
  • ✅ Custom error metadata (via Zod)
  • ✅ Type safety for custom assertions (may require a significant refactor)
• ✅ Lean on Zod more for builtin assertion implementations and use its error-reporting facilities
• Keypaths / property drilling
• Boolean logic syntax (and, or, not) while still avoiding chainable APIs. Though I might convince myself that chainable APIs are OK for this and this only.

Maybe Later

• Assertions for all Zod v4 builtins, essentially (which suggests some dynamic generation of assertions)
• Lower-level plugin API for those things which are more involved than custom assertions
• Basic spies (rationale: it is v. common to want to just check that a function gets called)
  • Draw a line in the sand: no stubs, no mocks, no fakes
  • Simple function instrumentation via Proxy or something; a way to check if a function was called, how many times, and with what
  • Adapters which generate assertions from 3p spy providers?
• Is snapshot testing the responsibility of an assertion library?
• Drop hard dependency on Node.js and expand support to other environments; as of this writing there are only three (3) bits used from Node.js:
  • node:test (dev)
  • node:util.inspect (dev)
  • node:assert.AssertionError
• Pull in multiple test frameworks into the dev env and assert basic compatibility with all of them
• Codemods for migration
• Support other object validation libraries (or wasn't there one that provided a translation layer?)
• See if there's some way to leverage asserts type predicates without too much boilerplate & tedium. Maybe make a TS feature request to allow asserts and/or is keyword to be used in types (or find out why they didn't want to do that).