Sync Upstream logo

Sync Upstream

OrganizationPopular
nyaruka
sync-upstream

Sync this Go port with a new upstream google/libphonenumber release — regenerate the embedded metadata and reconcile the ported Java logic. Takes an optional argument — `data` (metadata regen only), `code` (code reconciliation only), or empty for both. Use when the user says "sync with upstream", "do an upstream sync", "reconcile against libphonenumber", "pull the latest libphonenumber", bump to a new libphonenumber version, or invokes /sync-upstream. SYNC.md holds the state (baseline version + deliberate divergences); this skill is the procedure.

Overview

Publishernyaruka
Repositoryphonenumbers
Skill namesync-upstream
Stars
1.6K
Forks
183
Bundled files
Instructions only
LicenseMIT
Links
  • Markdown instructions

    A SKILL.md file the model loads on demand, so it only costs tokens when a request actually matches.

  • Works with any LLM

    AI skills are plain Markdown, not provider-specific code, so this works with GPT, Claude, Gemini, Grok, or a local model.

  • Self-contained

    Everything the model needs lives in the instructions — no extra files to sync.

  • Open source

    Published by nyaruka on GitHub. Read the source before you install it.

Installation

Install the Sync Upstream AI skill in TypingMind to use it with any LLM, or drop it into another agent that reads SKILL.md.

1

Install in TypingMind

TypingMind installs a skill straight from its GitHub folder — it reads SKILL.md, bundles the resource files, and stores the result locally.

  1. Open the app and go to Plugins → Skills.
  2. Choose "Install from GitHub".
  3. Paste the skill folder URL below and confirm.
    https://github.com/nyaruka/phonenumbers/tree/main/.claude/skills/sync-upstream
  4. Enable the skill in any chat where you want it available.
Plugins → Skills → Add skill → From GitHub URL, then paste the folder URL and press Continue.
2

Install in another agent

Any agent that reads the Agent Skills format can use this skill — copy the folder into that agent's skills directory.

Claude Code — .claude/skills
git clone --depth 1 https://github.com/nyaruka/phonenumbers.git /tmp/phonenumbers
mkdir -p .claude/skills
cp -r /tmp/phonenumbers/.claude/skills/sync-upstream .claude/skills/sync-upstream
Restart Claude Code after copying so it picks up the new skill.

Use it in TypingMind

Enable Sync Upstream in any TypingMind chat and the model takes it from there. Its name and description sit in the system prompt, and the moment a request matches, the model loads the full instructions itself — you never invoke it by hand, and it costs no tokens until it is actually used.

The model loads Sync Upstream on its own as soon as a request matches it.

Works with any AI model

AI skills are plain Markdown instructions rather than provider-specific code, so Sync Upstream is not tied to the model it was written for. Install it once in TypingMind and use it with GPT-5, Claude, Gemini, Grok, DeepSeek, Mistral, Llama, or a local model you run yourself — all on your own API keys.

  • Loaded only when it is needed

    The system prompt carries just the name and description. The instructions are fetched on the first matching request, so an idle skill costs nothing.

  • Switch models mid-chat

    Because the skill is instructions rather than code, changing model does not break it — the next model reads the same SKILL.md.

Skill instructions

This is the SKILL.md content the model loads. Read it before installing — a skill is instructions your model will follow.

Sync with upstream libphonenumber

This package is a Go port of google/libphonenumber, tracking its Java reference implementation. A sync is two independent operations — do them in order, but understand they're different in kind:

  1. Metadata regen — mechanical. Rebuild the embedded data/ blobs from a new upstream release. Fully automated by cmd/buildmetadata.
  2. Code reconciliation — judgment. Port new Java logic changes into the Go files. Can't be automated; this is the real reason SYNC.md exists.

State lives in SYNC.md: the code baseline (Code reconciled against vX.Y.Z) and the deliberate divergences (what we intentionally do not port). Read it first. The embedded-metadata version is tracked separately in the generated metadata/version.go (metadata.Version) — never hand-edit that.

Run everything from the repo root (/Users/rowan/nyaruka/phonenumbers).

Argument

The skill takes one optional argument selecting which operation(s) to run:

  • /sync-upstream data — Phase 1 only (metadata regen). An automated metadata update can invoke this mode directly.
  • /sync-upstream code — Phase 2 only (code reconciliation, against the metadata version already built — see the note at the top of Phase 2).
  • /sync-upstream — both phases, in order.

Whichever mode runs, finish with Finalize below: commit the changes and stop. This skill never releases — no tag, no push, no CHANGELOG.md. The user cuts releases separately with /release.


Phase 1 — Metadata regen (mechanical)

sh
go run ./cmd/buildmetadata            # latest upstream release
# go run ./cmd/buildmetadata v9.0.32  # or pin a specific tag

This resolves the target tag, wipes and re-clones upstream into _build/ (a shallow --depth=1 checkout at the target tag — gitignored), rebuilds every embedded blob (data/, metadata/data/, carrier/data/, geocoding/data/, timezone/data/), and rewrites the generated metadata/version.go with the tag it built from.

Then:

sh
go test ./...

Review the diff. The only files that may change are the embedded blobs (data/, metadata/data/, carrier/data/, geocoding/data/, timezone/data/) and the generated metadata/version.go:

  • No changes at all → the metadata is already up to date; nothing to commit for Phase 1. In data mode report that and stop; in a full sync continue to Phase 2 (the code baseline may still lag the metadata version).
  • Changes outside that set → something is wrong; stop and report the unexpected changes without committing.

In data mode, that's the whole job — skip to Finalize. Phase 1 has already bumped metadata/version.go, and the Code reconciled against baseline in SYNC.md stays where it is (the two versions are tracked separately for exactly this case).

The target tag you built becomes the new code baseline once Phase 2 reconciles against it.


Phase 2 — Code reconciliation (judgment)

Goal: apply any upstream Java logic changes between the old baseline (SYNC.md) and the target to each Go port. The target is the tag Phase 1 built — in code-only mode, the version already in metadata/version.go (metadata.Version), i.e. reconcile the code up to where the embedded metadata already is.

Get both tags side by side for diffing

Phase 1 leaves _build/ as a shallow clone at the target tag. In code-only mode _build/ may be missing (it's gitignored) — recreate it at the target first:

sh
TARGET=v9.0.32    # metadata.Version, from metadata/version.go
git clone --depth=1 --branch "$TARGET" https://github.com/google/libphonenumber _build

Then fetch the baseline tag into it so you can diff tree-to-tree (no merge base needed — git diff A B compares trees):

sh
BASE=v9.0.31      # from SYNC.md "Code reconciled against"
git -C _build fetch --depth=1 origin tag "$BASE"

Diff each ported source, baseline→target

The per-file headers are the source of truth for the Go→Java mapping — every reconcilable file carries a // Port of <upstream/path>. header. Enumerate them:

sh
grep -rn "Port of" --include='*.go' . | grep -v _test.go | grep -v _build/

For each upstream path P, diff it across the window and reconcile anything non-trivial:

sh
git -C _build diff "$BASE" "$TARGET" -- "$P"

The mapping (kept current; if files move, the headers win and this table is refreshed):

Goupstream Java path
phonenumberutil.go, enums.gojava/libphonenumber/src/com/google/i18n/phonenumbers/PhoneNumberUtil.java
asyoutypeformatter.gojava/libphonenumber/src/com/google/i18n/phonenumbers/AsYouTypeFormatter.java
shortnumberinfo.gojava/libphonenumber/src/com/google/i18n/phonenumbers/ShortNumberInfo.java
phonenumbermatcher.gojava/libphonenumber/src/com/google/i18n/phonenumbers/PhoneNumberMatcher.java
phonenumbermatch.gojava/libphonenumber/src/com/google/i18n/phonenumbers/PhoneNumberMatch.java
errors.gojava/libphonenumber/src/com/google/i18n/phonenumbers/NumberParseException.java
internal/regexcache/java/libphonenumber/src/com/google/i18n/phonenumbers/internal/RegexCache.java
internal/regexbasedmatcher/java/libphonenumber/src/com/google/i18n/phonenumbers/internal/RegexBasedMatcher.java, .../internal/MatcherApi.java
metadata/source.go, alternateformats.gojava/libphonenumber/src/com/google/i18n/phonenumbers/metadata/source/*, .../MetadataLoader.java
carrier/java/carrier/src/com/google/i18n/phonenumbers/PhoneNumberToCarrierMapper.java
geocoding/java/geocoder/src/com/google/i18n/phonenumbers/geocoding/PhoneNumberOfflineGeocoder.java
timezone/java/geocoder/src/com/google/i18n/phonenumbers/PhoneNumberToTimeZonesMapper.java
internal/prefixmapper/java/internal/prefixmapper/src/com/google/i18n/phonenumbers/prefixmapper/*
internal/metadatabuilder/tools/java/common/src/com/google/i18n/phonenumbers/BuildMetadataFromXml.java

internal/serialize and internal/stringbuilder are Go-specific scaffolding (gzip/binary decode; a java.lang.StringBuilder stand-in) with no upstream logic to reconcile — skip them.

The corresponding upstream tests move in lockstep. The Go tests mirror the Java test classes (e.g. PhoneNumberUtilTest → phonenumberutil_test.go and friends); diff those too and port new cases:

sh
git -C _build diff "$BASE" "$TARGET" -- "$(echo "$P" | sed 's#/src/#/test/#')"

Reconcile

For each non-empty diff, translate the relevant change into the Go port. While doing so:

  • Preserve upstream source order. The core ports keep their functions/methods in the same order as the upstream Java so the next sync's diff lines up — maintain that as you reconcile. A few files are deliberately reorganized for Go idiom and don't follow upstream order (metadata/source.go, internal/regexcache, internal/regexbasedmatcher, internal/prefixmapper); that's expected.
  • Java-shaped over idiomatic — in ports. Inside a // Port of … file, keep constructs close to the Java so the next diff lines up, even when a linter (gopls modernize, staticcheck) suggests a more idiomatic Go form — e.g. don't rewrite an index for loop as for i := range n, or a manual scan as slices.Contains. Let those modernizations ride in with upstream when the Java itself changes. Non-ported code (cmd/, CI, build glue, and Go-specific scaffolding like internal/serialize / internal/stringbuilder) has no Java counterpart to track, so make it fully idiomatic.
  • Honor deliberate divergences. Check SYNC.md before porting something back; some upstream code is intentionally absent (e.g. Mockito-style metadata-source injection tests, Java iterator semantics with no Go equivalent). Don't re-introduce it. If you make a new intentional divergence, record it.
  • Keep it a port, not an enhancement. Match libphonenumber's behaviour; don't add API.
  • Mirror Java's visibility. A symbol that is public in Java is exported in Go; one that is private/package-private/@VisibleForTesting stays unexported (lower-case camelCase). Constants especially: REGION_CODE_FOR_NON_GEO_ENTITY is the only public static final upstream, so it is the lone exported constant — every other ported regex/char-class/length-limit/mapping is unexported. The deliberate exceptions are the Go-idiomatic Err* vars (standing in for Java's checked exceptions) and the public enum types. Don't export a new internal just because it's a package-level const/var.
  • Java overloads → distinct Go names. Go can't overload, so a set of Java methods sharing a name maps to several Go funcs: give one the bare PascalCase name and suffix the rest with the parameter(s) that distinguish them. Follow the established pattern — Parse/ParseToNumber, IsNumberMatch/IsNumberMatchWithNumbers/IsNumberMatchWithOneNumber, GetExampleNumberForType/GetExampleNumberForTypeInRegion, IsPossibleNumber/IsPossibleNumberFromRegion, IsNumberGeographical/IsNumberGeographicalForType. Porting an overload that already exists upstream is faithful porting, not the "don't add API" case above — fill the gap rather than skipping it. Which overload keeps the bare name (and the exact suffix) is a judgment call; flag new public names in the PR.
  • Don't hand-edit generated files (metadata/version.go, *.pb.go) or the data/ blobs — those come from Phase 1 / protoc.

Then make it green:

sh
go test ./...

Finalize

If code was reconciled (Phase 2 ran), update SYNC.md (state only):

  1. Bump the Code reconciled against version to the target tag.
  2. If anything new was intentionally left unported, add it under Deliberate divergences.

Then, with go test ./... green, commit the changes:

  • Metadata only (data mode): commit the blobs and metadata/version.go with the message Updated metadata to <tag>, where <tag> is the upstream tag it was built from (metadata.Version in metadata/version.go), e.g. Updated metadata to v9.0.33.
  • Code reconciliation: record what was reconciled (and the version) in the commit message — git history is the sync log, so SYNC.md doesn't keep one.

Stop there. Do not push, tag, or touch CHANGELOG.md — those belong to the release process. Finish by reporting that the changes are committed and a release can be cut with /release.

Leave public-facing text (commit messages, comments) describing the software in general terms.

Frequently asked questions

What does the Sync Upstream AI skill do?

Sync this Go port with a new upstream google/libphonenumber release — regenerate the embedded metadata and reconcile the ported Java logic. Takes an optional argument — `data` (metadata regen only), `code` (code reconciliation only), or empty for both. Use when the user says "sync with upstream", "do an upstream sync", "reconcile against libphonenumber", "pull the latest libphonenumber", bump to a new libphonenumber version, or invokes /sync-upstream. SYNC.md holds the state (baseline version + deliberate divergences); this skill is the procedure.

Why use Sync Upstream on TypingMind?

Because you install it once and use it with any model. Sync Upstream is plain Markdown rather than provider-specific code, so the same skill runs on GPT-5, Claude, Gemini, Grok, or a local model — and you can switch model mid-chat without it breaking. TypingMind runs on your own API keys, so you pay providers directly instead of a per-seat subscription, and your skills and chats stay in your own storage.

How do I install Sync Upstream in TypingMind?

Open Plugins → Skills → Install from GitHub in TypingMind and paste https://github.com/nyaruka/phonenumbers/tree/main/.claude/skills/sync-upstream. TypingMind reads its SKILL.md and installs it as a skill you can enable per chat.

Which AI models can use Sync Upstream?

Any model you connect in TypingMind. AI skills are plain Markdown instructions rather than provider-specific code, so GPT, Claude, Gemini, Grok, and local models can all load this skill when a request matches it.

How many AI models can I use with Sync Upstream?

As many as you like. As long as a model supports skills, you can use Sync Upstream with it — GPT, Claude, Gemini, Grok, DeepSeek, Mistral, Llama and more — all on TypingMind with your own API keys.

Is the Sync Upstream AI skill free?

Yes. It is published on GitHub by nyaruka under the MIT license. You only pay your own AI provider for the tokens you use.

What are AI skills?

An AI skill is a reusable instruction bundle that teaches an AI model how to do one specific task. It follows the open Agent Skills format: a SKILL.md file with a name and description, plus any scripts, templates or reference files the model may need. The model reads the instructions only when your request matches the skill, so an installed skill costs nothing until it is used.

How are AI skills different from plugins or MCP servers?

A plugin or MCP server gives a model new tools to call — code that runs somewhere and returns a result. An AI skill gives the model knowledge and process instead: how to approach a task, which steps to follow, what good output looks like. Skills are plain Markdown, so they need no server, no API key and no runtime, and they work with any model.

View all

Set up your own AI workspace now

Get notified about new features and future giveaways by subscribing to our newsletter 👇