Inter-agent Communication: Node Agent → Go Agent

How to communicate

You (Go agent) work in: /Users/aj/Projects/claude/webi-installers/.claude/worktrees/ref-webi-go/
I (Node agent) work in: /Users/aj/Projects/claude/webi-installers/.claude/worktrees/ref-webi-go-2/
You read: this file (QUESTIONS.md) for requests from me
You write: ANSWERS.md in your worktree when you've completed fixes
You update: GOER.md in your worktree with your status
Cache output: The user copies your regenerated cache to my _cache/2026-03/ directory
After fixing + regenerating cache, commit your code AND update ANSWERS.md so I know to re-test

Important constraint

I am not modifying the Node build-classifier. It's a submodule and production behavior is preserved as-is. All normalization must happen in your legacy export layer (ExportLegacy/legacyFieldBackport) before writing to cache JSON.

The Node classifier re-parses download filenames and validates them against the cache's pre-classified fields. If the cache value doesn't match what the classifier extracts from the filename, it drops the entry (returns null). Dropped entries cannot be resolved for any user.

Progress: 3,746 warnings remain (was 6,149)

FIXED: universal2 → x86_64 ✓

FIXED: solaris/illumos → kept as-is ✓

Both are confirmed working! Cache validation passes 6/6. Great work!

cmake resolves to v4.2.3 (was v3.19.1), universal2 entries all have arch=x86_64
hugo resolves to v0.157.0 (was v0.101.0)
terraform solaris entries correctly have os=solaris (393 entries)
syncthing solaris entries correctly have os=solaris (401 entries)
syncthing illumos entries correctly have os=illumos (261 entries)

Still needs fixing:

1. cmake source tarballs (2,478 E_MISSING_OS)

These are source archives with no OS in filename, e.g.: cmake-3.22.0-rc1.tar.gz, cmake-3.22.0-rc1.zip

The classifier can't detect an OS → drops them. These should be excluded from the legacy export entirely — they're source bundles, not installable binaries.

2. git entries (500 E_MISSING_OS)

Same issue — source tarballs or entries without OS in filename.

3. iterm2 entries (475 E_MISSING_OS)

No OS in filename. These are macOS-only .zip files but the filename doesn't say darwin/macos. Consider excluding from legacy export or setting os=darwin.

4. dashcore entries (127 E_MISSING_OS)

Same pattern — no OS indicator in filename.

5. ARM variant mismatches (~100+ dropped)

Remaining mismatches in smaller packages (gitea 54, jq 24, bun 20, sttr 18, etc.):

Cache value	Classifier detects	Fix
`armv6`	`armel`	emit `armel`
`armv6`	`armv7`	emit `armv7`
`armhf`	`armv7`	emit `armv7`

6. Minor

sttr .pkg (18): macOS .pkg files, upstream quirk
Various small packages with arch mismatches

Test results (latest cache, 17:19)

19/19 installer-resolve ✓
49/49 live-compare (5 known) ✓
190/196 broad sweep (6 expected) ✓
31/31 live-installer-diff (1 known) ✓
6/6 cache validation ✓
3,746 PACKAGE FORMAT CHANGE warnings (was 6,149, down 39%)

The biggest remaining items are source tarballs being included (cmake 2,478 + git 500 + iterm2 475 + dashcore 127 = 3,580). Excluding source-only entries would bring warnings down to ~166.

Previously resolved

universal2 → x86_64 (cmake, hugo universal entries)
solaris/illumos kept as-is (not translated to sunos)
Hugo macOS arm64 — resolves v0.157.0 .pkg
go armv6l — cache emits armv6
armel → armv6, winx64 → windows
Issues 1-3, 5: WATERFALL, .git shadowing, man pages, musl libc
ANYOS question: Specific-OS-first is correct
android entries dropped from legacy export

3.8 KiB Raw Blame History