Skip to main content

Unredact tool

The Unredact tool reverses redact: it restores the original values into the redaction placeholders across a block's source and target runs, then removes the in-process secret annotation so nothing sensitive reaches the writer. It is the closing half of the redact → process → unredact cycle, run after the protected content has been processed (for example, translated).

Restoration uses the recovery record left by redact. By default it reads the in-process secret annotation on the block; in external mode it reads originals from a sidecar vault (used by the extractmerge workflow). Both an ID-based restore (for structure-preserving formats and inline codes) and a text-based restore (for formats that flattened a placeholder to its visible string on write) are applied, so placeholders are recovered in either case. The vault path is supplied programmatically by the caller rather than as a schema field, so this tool exposes no schema parameters.

IDunredact
SourceBuilt-in
Categorytext-processing
Cardinalitymonolingual
Tagssecurity, redaction

This tool has no configurable parameters.

Processing notes

  • A transformer that restores originals into both source and target runs — it returns an edit plan and the framework applier rewrites the block, rebasing surviving source overlays across the restore.

  • Reads the in-process secret annotation by default, or a sidecar vault in external mode.

  • Removes the secret annotation after restoring, so sensitive originals never reach the writer.

Limitations

  • Exposes no schema parameters; the optional sidecar vault path is supplied programmatically by callers (e.g. extract → merge).

  • Restores only placeholders recorded by a preceding redact run; there is nothing to restore without that recovery record.

← Back to the Tool Reference