← Back to journal

A Captured Import Is Not Browser Resolution Authority

Why module instances and import maps remain a separate question from dependency presence.

Evidence basisKalu, Korpus, and frozen Ushman source/docs snapshot reviewed September 17, 2026; no new runtime qualification performed
DisclosureEngineering or methodology note based on a September 17, 2026 snapshot. Source-level behavior is not a public availability or universal compatibility claim. Any scenarios are illustrative, not customer outcomes.
Published by Kalu KodeRequest a KodeBack assessment →

Scope: Engineering or methodology note based on a September 17, 2026 snapshot. Source-level behavior is not a public availability or universal compatibility claim. Any scenarios are illustrative, not customer outcomes.

Seeing an import in captured JavaScript can tell us that a dependency relationship exists. It does not necessarily tell us how the browser resolved that import, which module instance was shared, or which realm owned the result.

That distinction is central to a current KodeBack limitation. Captured module imports do not automatically supply effective browser-resolution authority for externalization. The source reports that gap explicitly rather than treating dependency presence as permission to rewrite every consumer.

The same name can hide different execution relationships

A synthetic application might load related modules through different documents, import maps, or execution contexts. Two specifiers that look similar in text do not automatically establish one shared runtime instance. Conversely, different-looking specifiers may resolve into relationships that matter to singleton behavior.

Replacing those imports without understanding the effective resolution can change identity-sensitive behavior even when each replacement module exports familiar names.

Preserve the unresolved field

The current fingerprint path can complete on useful presence or revision evidence while module-resolution authority remains unavailable. The diagnostic FINGERPRINT_CAPTURED_MODULE_RESOLUTION_UNAVAILABLE should not be read as “nothing was learned.” It means a particular later claim has not been established.

That is why a report can be valuable without making the chunk externalizable. Evidence should guide the next step without acquiring an authority it did not produce.

Imported package publication has its own closure

Where the current direct externalization path publishes an import map, that map is derived from certified roots and authenticated distribution modules. Its verifier walks the supported static closure and rejects missing, extra, or misdirected entries. Unreached dynamic imports do not receive invented mappings or network allowances.

These are bounded integration rules, not proof of arbitrary browser module equivalence. The captured-import limitation remains important outside the specific certified path.

For engineers evaluating recovery, ask how the tool knows that the proposed replacement preserves the relevant module-instance relationships. A list of matching import names is not a complete answer. Package identity and browser resolution are connected, but they need their own evidence before the connection becomes a safe rewrite.