chore(npm): migrate to @bytefolk/mem-mcp v0.1.2 (#153)

[sun-970]

Summary

• Rename @fullstack-ai-infra/mem-mcp@0.1.1 → @bytefolk/mem-mcp@0.1.2 across all npm, MCP metadata, cache path, and documentation surfaces.
• Bump monorepo version to 0.1.2 on every version surface (Go, Python worker, web, Helm chart, docs, CHANGELOG).
• Preserve the old @fullstack-ai-infra/mem-mcp@0.1.1 package as installable until migration is proven.

Changes

Area	What changed
npm/package.json	name → @bytefolk/mem-mcp, version → 0.1.2, mcpName → io.github.bytefolk/mem-mcp
npm/install.js	PACKAGE constant + all cache path segments → bytefolk
npm/platforms.js	Error message package reference → @bytefolk/mem-mcp
npm/server.json	mcpName → io.github.bytefolk/mem-mcp, version → 0.1.2
npm/README.md	Install commands, Claude Desktop/Code config, cache path docs
npm/*.test.js	Cache path assertions + version references updated
server/cmd/mem-mcp/main.go	Version → 0.1.2, comment → npm/@bytefolk/mem-mcp
scripts/validate_release_version.sh	Comment reference → npm/@bytefolk/mem-mcp
README.md	Smithery badge → @bytefolk/mem-mcp
docs/maintainers/releasing.md	npm verify/publish references → @bytefolk/mem-mcp
CHANGELOG.md	New [0.1.2] - 2026-09-02 section with migration entry
Helm, web, worker, docs	All version fields → 0.1.2

Test plan

• npm test — 34/34 pass (1 Windows-only skip)
• scripts/validate_release_version.sh 0.1.2 — PASS
• Clean install on Linux/macOS/Windows: npm install @bytefolk/mem-mcp
• npm readback: access, metadata, provenance, Trusted Publisher, latest

Closes #153

[waterbro-8]

Closing under the fork-workflow decision recorded on 2026-09-03: repository
automation is not being enabled for fork pull requests, so a fork head cannot
carry a CI result, and every acceptance gate in this repository is written
against checks that ran. Nothing in this comment is a judgment that the work is
wrong; where it is right, it is re-landed on an organization branch instead.

#153. Two of the three things this branch does are decided differently as of
2026-09-03, and the third is re-landed as #162.

It touches 15 files. Only two lines of it are wanted now:

• wanted — mcpName → io.github.bytefolk/mem-mcp in npm/server.json and
  npm/package.json. That is exactly what fix(npm): follow the registry identifier to the bytefolk namespace #162 does, on its own, plus a test
  that derives the expected namespace from the coordinate npm/install.js
  downloads from.
• not wanted — the package rename to @bytefolk/mem-mcp. The decision keeps
  the published scope, so the Helm chart values, npm/install.js, the installer
  tests, platforms.js, npm/README.md and the deployment docs all stay as they
  are. Existing "args": ["-y", "@fullstack-ai-infra/mem-mcp"] configs and cache
  paths keep working.
• not wanted here — the 0.1.2 version bump. A version belongs to the
  release commit; scripts/validate_release_version.sh:39 pins
  npm/server.json's version to the tag, so bumping it in a feature branch is
  the wrong place even when a release is planned.

A correction to my own reading, which is what led here: my #104 comment today
argued that the coordinate and #153 had to move in one action, because doing
them separately would publish a registry record carrying the old coordinate.
That conflated two namespaces. The registry identifier's namespace comes from
the repository owner; the npm package's scope is an unrelated string. Only the
first had to move, and #162 moves it alone.

Since #153 still reads "Migrate mem-mcp to @bytefolk/mem-mcp v0.1.2", it now
contradicts the decision it is tagged status:ready under. It should be
rewritten before someone implements it again.

The commits are not lost. A closed fork PR keeps its head ref:

git fetch https://github.com/bytefolk/mem.git refs/pull/154/head:pr-154

Every file in this branch was therefore available to the re-doing work, whether
or not it was used.