Skip to main content
← Back to list
01Issue
FeatureOpenSwamp CLIPublic
AssigneesNone

Relationships

#3991 Move the Zod argument-field extractor into @swamp-club/extension-scoring so the CLI and registry read schemas alike

Opened by skunk-ape · 10/26/2026

The CLI (src/domain/extensions/extension_content_extractor.ts) and the registry (swamp-club lib/app/extract-extension-content.ts) each keep their own copy of the static Zod field parser that produces content-metadata arguments, including the sensitive flag. Both test MARKED_SENSITIVE from the shared package, but field splitting, name parsing and kind coverage are separate code: the registry copy does not extract the extensions kind at all. Once the registry scores secret-safe-docs from a server re-read of the archive (swamp-club#3863), any difference between the two parsers is a CLI/registry scoring parity gap. Move the parser into the package, have both sides call it, and add vectors for it.

02Bog Flow
◉OPEN○TRIAGED○IN PROGRESS○SHIPPED

Open

10/26/2026, 10:11:51 PM

No activity in this phase yet.

03Sludge Pulse
Editable. Press Enter to edit.

skunk-ape commented 10/26/2026, 10:23:58 PM

Measured while working swamp-club#3863: across 680 swamp-extensions packages whose models mark an argument sensitive, the registry's extractor finds every marked name in 667, some of them in 9, and none in 4. The 4 keep their schemas in models/_lib files, which neither extractor resolves (both read one source file per entry point), so those agree. The 9 partial matches are field-parsing differences between the two copies, and they are what this issue would close.

Sign in to post a ripple.