[ruvnet]

Why

Only four @claude-flow/* components are bundled into the release tarballs — INTERNAL_RUNTIME_PACKAGES in stage-internal-runtime-bundles.mjs: security, codex, mcp, plugin-agent-federation. Everything else the CLI depends on — cli-core, neural, shared, memory — resolves from the registry at whatever v3/@claude-flow/cli/package.json pins.

So a merged source change in one of those does not ship with the three-package train. The release silently carries the last published copy of that leaf.

This has bitten twice:

• SONA RUFLO_INTELLIGENCE_MODE (#3334) never reaches consumers — @claude-flow/memory leaf package not republished #3335 — feat(sona): default the learning mode from RUFLO_INTELLIGENCE_MODE #3334 fixed RUFLO_INTELLIGENCE_MODE in @claude-flow/memory. v3.42.1 shipped cli/claude-flow/ruflo only; memory was last published weeks earlier, so a fresh consumer install had zero references to the fix.
• fix(memory): activate reasoningBank — wire the real embedder + correct the bridge contract (#3327 Finding A) #3390 (reasoningBank/vectorBackend report enabled:true but never activate (bridge-fallback/substring only); smart:true reports an RRF score mislabeled as cosine similarity #3327 Finding A) — the reasoningBank activation fix spanned cli/src/memory/memory-bridge.ts and memory/src/controller-registry.ts. Publishing only the train would have shipped the bridge half and left the embedder half unreachable, leaving reasoningBank exactly as inert as before. Caught by hand; memory@3.0.0-alpha.25 went out alongside v3.42.5.

What makes it easy to miss: CI goes green. The leaf's own package tests run against workspace source, so the change looks verified while being unshippable.

What the guard asserts

1. Every non-bundled leaf's workspace version is covered by the CLI's declared range. A miss means the release resolves to older code than this repo's source.
2. (diff mode) If a leaf's publishable source changed vs the base ref but its version did not — fail, with remediation.

Design choices that matter:

• Imports INTERNAL_RUNTIME_PACKAGES instead of restating it, so the bundled/non-bundled sets cannot drift apart. A hardcoded copy would go stale the moment someone adds a package to the staging script — and then wave through the exact bug this exists to catch.
• Exits non-zero if it inspects zero leaves, so a broken classifier can't report clean.
• A skipped diff check reports NOT EXERCISED, never a silent pass.
• Test/fixture/docs changes in a leaf don't trip it — they can't alter what consumers receive.

Accepted finding

@claude-flow/shared is entered as a time-boxed waiver (expires 2026-10-21): the CLI pins 3.0.0-alpha.7 while the workspace is 3.0.0-alpha.8, which is published — stale, not unshippable. Raising the pin edits v3/@claude-flow/cli deps and therefore forces a pnpm-lock.yaml regen in the same change (#2540 → #2552 — frozen-lockfile otherwise fails ~25 jobs), so it wants its own PR. An expired entry fails the guard rather than muting it.

Negative controls

A guard that cannot fail is worthless, so all four paths were exercised:

scenario	expected	result
memory source changed, no version bump	FAIL	✅ FAIL — replays #3390
same change with a version bump	pass	✅ pass
test-only change in a leaf	pass	✅ pass
waiver expiry backdated to 2020	FAIL	✅ FAIL — "no longer honoured"

🤖 Generated with Claude Code