Refactor transactionsStructural proofRuntime characterizationEquivalencePerformance
KodeProofis a private-alpha research system. Its verdicts are scoped empirical evidence, not formal proof, and remain one part of engineering review.

Async Side Effects Need Provenance

An effect observed after an await should be attributed only when the harness owns the continuation. Otherwise the honest result is no claim.

Equivalence Is Tiered, Not Binary

Static structure, module loading, runtime traces, visual states, and end-to-end flows cover different failure classes. One green tier cannot stand in for the others.

Initialization Order Is Behavior

Moving code can preserve every exported value and still change effects, cycles, registries, scheduling, or top-level-await order.

Module-Boundary Recovery Should Be Willing to Abstain

Recovered bundles often lack original file boundaries. A structural system should use exact evidence where it exists, deterministic fallbacks where it does not, and refuse contradictory paths.

Performance Refactors Need Parity Evidence

A faster benchmark is not enough if caching, parsing, batching, or concurrency changes values, ordering, resource ownership, or user-visible behavior.

Preserve Public Entrypoints Without Leaving Compatibility Facades

A hard-cut refactor should move consumers to the new owner, prove the retained contract, and refuse when closure requires a permanent re-export shell.

A Refactor Is a Transaction, Not a Diff

Risky changes become governable when intent, before-state, allowed mutations, evidence, verdict, and transfer are one bound transaction.

Relocating a File Is a Semantic Change

Framework scanners, routes, singleton identity, provider scope, styles, assets, workers, and generated consumers can all make a filesystem move observable.

Runtime Characterization Catches What Static Analysis Cannot See

Captured calls, receivers, values, effects, readiness, and state coverage can expose refactor regressions beyond the source graph—when their limits stay explicit.

From Changed Symbols to the Tests That Can See Them

A workspace symbol map can connect bindings, imports, reverse paths, and focused tests—but incomplete discovery must remain visible.

A Verification Harness Must Test Its Own Ability to Fail

Baseline checks, calibrated mutations, differential oracles, false-block reporting, and harness-failure categories keep green verdicts from becoming self-certification.