Agent skill

Mlkit Release

by d-date in d-date/google-mlkit-swiftpm

A skill your agent uses when triaging incoming issues or PRs for google-mlkit-swiftpm, or when cutting a release.

Apache-2.0Auto-check passedMobile

Install Mlkit Release

skills CLI
$ npx skills add d-date/google-mlkit-swiftpm --skill mlkit-release -a claude-code

Project install by default; add -g for ~/.claude/skills/.

GitHub CLI
$ gh skill install d-date/google-mlkit-swiftpm mlkit-release --agent claude-code

Project scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).

Manual copy
$ git clone --depth 1 https://github.com/d-date/google-mlkit-swiftpm.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.claude/skills/mlkit-release .claude/skills/mlkit-release && rm -rf skills-src

Use ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.

Claude Code skills documentation · loads skills from .claude/skills/

Facts

Skill name
mlkit-release
GitHub stars
148
Token cost
~1.7k tokens
SKILL.md length
778 words
Files
1
Skills in repo
1
Repo updated
First seen
Licence
Apache-2.0

At a glance

A skill your agent uses when triaging incoming issues or PRs for google-mlkit-swiftpm, or when cutting a release.

  • Triaging incoming issues
  • SKILL.md covers Triage: match the report to a…, Triage: dependency PRs, Fix, then verify before… and Publish, plus 1 more section
  • Calls ruby, make and gh
  • PRs for google-mlkit-swiftpm

What it does

Mlkit Release is an agent skill from d-date/google-mlkit-swiftpm. Use when triaging incoming issues or PRs for google-mlkit-swiftpm, or when cutting a release. Covers the four failure modes consumers keep reporting (undefined symbols, "Invalid model path.", ITMS-91065, no arm64 simulator), how to tell them apart, and the build → verify → publish sequence.

Its SKILL.md is about 1.7k tokens, which your agent loads only when the skill is triggered. It is a single SKILL.md file with no bundled scripts.

It sits in Mobile, covering iOS development. It works with Swift, iOS and Xcode. The repository describes itself as: Google MLKit for iOS in Swift Package Manager. The licence is Apache-2.0.

When your agent uses it

  • Triaging incoming issues
  • PRs for google-mlkit-swiftpm
  • Cutting a release

Example prompts

  • “Invalid model path.”
  • “/mlkit-release”

What it can do on your machine

Read from SKILL.md and the folder at commit 4f3379b. It shows what the files ask for, not the result of running them.

  • Tool permissions

    Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.

    From allowed-tools in the SKILL.md frontmatter.

  • Runs code

    Shell commands in SKILL.md call:

    • ruby
    • make
    • gh
    • swift

    From the folder's file list and the shell code blocks in SKILL.md.

  • Network

    No URLs in SKILL.md. Its commands use gh, which can reach the network depending on how they are called.

    From URLs in SKILL.md, links to its own repository left out.

  • Credentials

    Names no API keys, tokens, secrets or passwords.

    From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.

Context cost

Mlkit Release loads about 1.7k tokens when it runs. Until then it costs about 76 tokens; SKILL.md has 778 words of instructions outside code blocks.

Always · name and description, kept in context so the agent knows when to use it
~76
When it runs · the whole SKILL.md, loaded when a task matches
~1.7k

Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.

Safety

Auto-check passed

The automated check found no risky patterns in SKILL.md.

Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.

SKILL.md

The full file from d-date/google-mlkit-swiftpm at commit 4f3379b, republished under its Apache-2.0 licence (© d-date). 778 words, ~1,692 tokens.

Download SKILL.mdSave it as .claude/skills/mlkit-release/SKILL.md (or your agent's skills folder).
name
mlkit-release
description
Use when triaging incoming issues or PRs for google-mlkit-swiftpm, or when cutting a release. Covers the four failure modes consumers keep reporting (undefined symbols, "Invalid model path.", ITMS-91065, no arm64 simulator), how to tell them apart, and the build → verify → publish sequence.

ML Kit SwiftPM triage and release

Read CLAUDE.md for the pipeline shape first. This skill is the procedure; that file is the reference.

Triage: match the report to a known mechanism

Almost every consumer report is one of these four. Identify the mechanism before changing anything — three of the four are invisible when building the Example app, because it links every product at once.

SymptomMechanismCheck with
Undefined symbols … _OBJC_CLASS_$_MLKXeno* when adopting one productThe .library product omits a transitive framework. A product links only the targets it lists.ruby scripts/check_product_closure.rb
MLKTextRecognizerInternalErrorCreationFailure "Invalid model path."The module's model bundle was never published. Pods/<module>/Resources/<Name>/ is the full list; ten of them exist only there, not nested in any frameworkls GoogleMLKit/*.bundle against ls -d Pods/*/Resources/*/, and ./scripts/verify_local_archive.sh
A framework binary that is ~33KB in the appThe framework binary is a bare Mach-O object, so Xcode relinks it as a dylib and dead-strips the datafile GoogleMLKit/<name>.xcframework/*/<name>.framework/<name> — must say ar archive
ITMS-91065: Missing signature on uploadA framework bundle on Apple's commonly-used-SDK list is embedded in the app. Xcode embeds a stub bundle for every static framework bundle it links../scripts/verify_local_archive.sh
Cannot build for the Simulator on Apple SiliconThe simulator slice has no arm64. Xcode 26 dropped Rosetta simulators, so an x86_64-only slice is unusable.lipo -archs on the simulator slice
Invalid MinimumOSVersion … is '' (90530) on uploadA Resources/*-Info.plist template is missing the key; the templates ship verbatim into consumers' appsruby scripts/verify_build.rb

Reports about missing dSYMs are expected and harmless: Google ships no DWARF, the frameworks are static, and their symbols land in the consumer's own dSYM. Close with that explanation.

Triage: dependency PRs

  • Package.swift dependency pins are generated from Podfile.lock by scripts/update_package_dependencies.rb. renovate.json disables the swift manager for the root Package.swift; close any PR that still slips through, and never merge one that breaks a pod constraint (e.g. GTMSessionFetcher is pinned < 4.0).
  • Ruby major upgrades are blocked on purpose: Ruby 4.0 broke CocoaPods on the macOS runners.
  • Anything touching Example/ only is safe to merge on a green CI run.

Fix, then verify before publishing

bash
make run                            # full pipeline into GoogleMLKit/
make verify                         # closure + artifacts + Package.swift parse
./scripts/verify_runtime.sh         # slices, Info.plists, symbol table
./scripts/verify_local_archive.sh   # runs ML Kit on the Simulator, archives for device
./scripts/verify_app_store_upload.sh --upload   # Apple-side validation and delivery

verify_local_archive.sh is the gate. It repoints Package.swift at the local XCFrameworks, runs ML Kit on the arm64 Simulator, archives for device, and fails if an ITMS-scanned SDK is embedded, if anything is linked dynamically, or if the app binary is too small to still contain the ML Kit model data. Never publish without a green run.

The runtime step is not optional theatre: a missing model bundle links and archives perfectly and only fails when inference runs. Two of the four recurring reports would have been caught years earlier by it.

verify_app_store_upload.sh --upload is what closes out ITMS-91065: Apple reports it during post-upload processing, so no local check can stand in for a real delivery. The App Store Connect side is already set up:

Show full SKILL.md (306 more words)Show less
  • app record SwiftPM Binary Verify (6814873277), bundle ID com.d-date.google-mlkit-swiftpm (7Y76Y69Z42), team N5U649DS4Z
  • provisioning profile SwiftPM Binary Verify App Store (MJ47UVM4SC), built from the Apple Distribution certificate JWTYWVHM8Z

Four things cost time the first time round; they are worth knowing:

  • Creating an app record needs a web session (asc web auth login); the public API cannot do it. The session defaults to whichever provider Apple returns first, which may be another organisation entirely -- check with asc web bundle-ids list before creating anything, and select the right one with --provider-id <numeric> --developer-team N5U649DS4Z (passing a wrong --public-provider-id conveniently lists the valid ones).
  • Automatic signing does not work here: Xcode has no signed-in account, so the profile has to be created through the API and installed with asc profiles local install.
  • Signing settings must not be passed to xcodebuild on the command line. They reach every target in the graph, and SwiftPM's resource-bundle targets fail with "does not support provisioning profiles". Put them in ExportOptions.plist instead.
  • altool wants its API key as a file in one of a few fixed directories, while asc auth login keeps it in the keychain. The script stages a copy and removes it afterwards.

To reproduce a consumer's single-product setup, the static closure check is enough — do not rely on the Example app.

Publish

bash
ruby scripts/update_version.rb <version>      # only when bumping the pod version
gh release create <version> --prerelease      # prerelease when the tag has a suffix
./scripts/upload_release.sh <version>         # uploads *.xcframework.zip and *.bundle.zip
ruby scripts/update_checksums.rb <version>    # rewrites Package.swift URLs + SHA256s
swift package dump-package > /dev/null

Prefer the Build MLKit XCFrameworks workflow for a normal version bump; do it locally only when the pipeline itself changed, since that is what needs verifying. Resource bundles must be uploaded too — consumers add them by hand, and scripts/download_bundles.sh fetches them from the release.

After publishing, reply on each issue the release fixes with the tag and the manual steps it still requires (bundles, -ObjC -all_load).

When adding a module

Podfile → Resources/<Name>-Info.plist → MLKIT_MODULES in the Makefile → Package.swift (.binaryTarget plus product wiring). Everything else derives from those. Finish with ruby scripts/check_product_closure.rb.

© d-date, Apache-2.0. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file

Files

Just SKILL.md in .claude/skills/mlkit-release of d-date/google-mlkit-swiftpm.

Open the folder on GitHubat commit 4f3379b

Compare with similar skills

Mlkit Release next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.

Mlkit Release compared with similar skills
SkillStarsUsed inTokensAuto-checkLicenceRepo updated
Mlkit Release this skilld-date/google-mlkit-swiftpm148—~1.7kAutomated safety check: PassApache-2.0
Inspector MCPipedro/Inspector170—~3.4kAutomated safety check: PassMIT
Dt Setup iOSDynatrace/dynatrace-for-ai161—~3.3kAutomated safety check: PassApache-2.0
Bootstrap iOSRayFernando1337/rayfernando-skills130—~1.1kAutomated safety check: PassApache-2.0
Swift iOS Standardsbladeofgod/flutter-ai-harness116—~375Automated safety check: PassMIT
Swift Patternssoftspark/ai-toolkit179—~3.1kAutomated safety check: PassApache-2.0

Similar skills

  • Inspector MCP

    ipedro/Inspector

    A skill your agent uses when an agent needs to inspect a live iOS app through the Inspector MCP bridge, register or troubleshoot InspectorMCPServer for a consumer Xcode project, or query, resolve…

    170 GitHub stars~3.4k tokensUpdated 5 mo ago
    MobileAuto-check passed
  • Dt Setup iOS

    Dynatrace/dynatrace-for-ai

    Set up the Dynatrace iOS SDK (OneAgent) in an iOS project using Swift Package Manager.

    161 GitHub stars~3.3k tokensUpdated 6 days ago
    MobileAuto-check passed
  • Bootstrap iOS

    RayFernando1337/rayfernando-skills

    Bootstrap agents for iOS, iPadOS, macOS, Swift, SwiftUI, SwiftData/Core Data, Swift Testing, Xcode build/test/debug, Simulator, App Intents, or XcodeBuildMCP work.

    130 GitHub stars~1.1k tokensUpdated 2 mo ago
    MobileAuto-check passed
  • Swift iOS Standards

    bladeofgod/flutter-ai-harness

    适用:编写或审查 iOS Swift Native Module、Bridge Adapter、Host 接线、Info.plist/Entitlements、SwiftPM 和原生 UI。不适用:Kotlin/Android、Dart Client 或 Wire Contract…

    116 GitHub stars~375 tokensUpdated 1 mo ago
    MobileAuto-check passed
  • Swift Patterns

    softspark/ai-toolkit

    Swift/iOS: SwiftUI, Combine, async/await, actors, SPM, Core Data, UIKit interop.

    179 GitHub stars~3.1k tokensUpdated today
    MobileAuto-check passed
  • Swift Development

    aiskillstore/marketplace

    Comprehensive Swift development for building, testing, and deploying iOS/macOS applications.

    430 GitHub stars~1.8k tokensUpdated today
    MobileAuto-check passed

Works with

Categories

Questions about Mlkit Release

What does Mlkit Release do?

A skill your agent uses when triaging incoming issues or PRs for google-mlkit-swiftpm, or when cutting a release. Mlkit Release is an agent skill from d-date/google-mlkit-swiftpm. Use when triaging incoming issues or PRs for google-mlkit-swiftpm, or when cutting a release.

When should I use Mlkit Release?

Mlkit Release fits situations like: triaging incoming issues; PRs for google-mlkit-swiftpm; cutting a release.

How do I install Mlkit Release in Claude Code?

Run `npx skills add d-date/google-mlkit-swiftpm --skill mlkit-release -a claude-code`. Or copy the skill folder (.claude/skills/mlkit-release in d-date/google-mlkit-swiftpm) into .claude/skills/mlkit-release in your project. Claude Code loads it when a task matches its description.

How do I install Mlkit Release in Codex?

Run `npx skills add d-date/google-mlkit-swiftpm --skill mlkit-release -a codex`. Or copy the skill folder (.claude/skills/mlkit-release in d-date/google-mlkit-swiftpm) into .agents/skills/mlkit-release in your project. Codex loads it when a task matches its description.

Can I use Mlkit Release in Cursor, Gemini CLI or GitHub Copilot?

Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add d-date/google-mlkit-swiftpm --skill mlkit-release -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/mlkit-release, .gemini/skills/mlkit-release, .github/skills/mlkit-release and .opencode/skills/mlkit-release in your project.

What does Mlkit Release need to run?

Going by SKILL.md and its folder, Mlkit Release needs the command-line tools its instructions call (ruby, make, gh and swift).

Does Mlkit Release access the network?

SKILL.md contains no URLs. Its commands use gh, which can reach the network depending on how they are called. This is read from the text; nothing was executed.

Is Mlkit Release safe to install?

Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.

What licence does Mlkit Release use?

Mlkit Release is published under the Apache-2.0 licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.

How many tokens does Mlkit Release use?

About 1.7k tokens (SKILL.md is roughly 6.8k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.

What are the alternatives to Mlkit Release?

Skills that share tags, products or a category with Mlkit Release: Inspector MCP (ipedro/Inspector, 170 stars), Dt Setup iOS (Dynatrace/dynatrace-for-ai, 161 stars), Bootstrap iOS (RayFernando1337/rayfernando-skills, 130 stars) and Swift iOS Standards (bladeofgod/flutter-ai-harness, 116 stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.

Who maintains Mlkit Release?

d-date (a GitHub user) maintains it in d-date/google-mlkit-swiftpm, which has 148 GitHub stars. The repository was last updated on September 24, 2026.

Source: d-date/google-mlkit-swiftpm on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.