background-assets-apple-hosted-packs
A Background Assets asset pack fails to download or EXTRACT on device or TestFlight — ProcessingPipelineError, AssetPackManagerError, STExtractionService faults, a pack stuck at downloading/failed, works dev-signed but fails on TestFlight (or vice versa), or external testers can't download despite READY_FOR_TESTING. Also for building/uploading Apple-hosted packs (ba-package, altool --upload-asset-pack, AssetPackManager) beyond what the base skill covers. Prerequisite: read ios-dev:background-assets-manifest-drift-blind-redownload first — it owns manifest-schema drift, the local content-cache check before re-downloading multi-GB sources, and pack-version immutability (new content = new pack id/version + client constant bump); this skill adds only the delivery/extraction sharp edges it lacks.
- License 许可
- MIT
- 最近更新
- 2026-11-07
- Stars
- ★ 1
- Forks
- 1
01 前言 Front Matter
- name
- background-assets-apple-hosted-packs
- description
- A Background Assets asset pack fails to download or EXTRACT on device or TestFlight — ProcessingPipelineError, AssetPackManagerError, STExtractionService faults, a pack stuck at downloading/failed, works dev-signed but fails on TestFlight (or vice versa), or external testers can't download despite READY_FOR_TESTING. Also for building/uploading Apple-hosted packs (ba-package, altool --upload-asset-pack, AssetPackManager) beyond what the base skill covers. Prerequisite: read ios-dev:background-assets-manifest-drift-blind-redownload first — it owns manifest-schema drift, the local content-cache check before re-downloading multi-GB sources, and pack-version immutability (new content = new pack id/version + client constant bump); this skill adds only the delivery/extraction sharp edges it lacks.
- promotion_target
- Fold into ios-dev:background-assets-manifest-drift-blind-redownload as an "extraction failures + install channels" section — this file carries only the delta.
03 SKILL.md 正文
Apple-hosted Background Assets: delivery & extraction sharp edges
Evidence base: Cubby BUG-038 (a 4-day multi-session investigation ending in a confirmed OS-version-specific platform regression), sessions 0051/0052.
1. Strip ALL dotfiles before packaging
ba-package's recursive directory selector has no exclude mechanism — every
hidden file in the source tree ships in the pack. Sweep for ALL dotfiles/hidden
directories, not just the obvious caches: Cubby excluded .cache/huggingface/ and
still shipped .gitattributes in the first "clean" rebuild. Verify with a listing of
the packaged contents, not the source dir.
2. Install channels resolve manifests differently
The BackgroundAssets framework distinguishes (verified via strings on the framework
binary) at least three manifest-resolution configurations: App Store (TestFlight),
App Review, and local-cache/development-override (dev-signed direct installs).
Consequence: a dev-signed repro proves nothing about TestFlight and vice versa —
Cubby saw AssetPackManagerError on dev installs but ProcessingPipelineError on the
TestFlight path for the same pack. Test the channel you ship through.
3. ASC has a separate external-testing approval state
xcrun altool --asset-pack-status reporting READY_FOR_TESTING is not the whole
story: App Store Connect tracks a distinct Ready-for-Internal vs Ready-for-External
state, and external testing needs its own submit/approval flow. Check the ASC web UI
(TestFlight → Asset Packs) — the CLI status alone doesn't confirm external testers can
download.
4. The error enums are private — capture Console.app live
ProcessingPipelineError, AssetPackManagerError and friends are private Apple enums;
their code numbers are not recoverable from public docs. The decisive diagnostic that
worked:
- Live Console.app, System → Errors and Faults, unfiltered, watching at the exact
failure moment (e.g.
STExtractionService.privilegedfaults,StreamingZipextraction errors). log collect --devicewas repeatedly unreliable ("Device not configured", regardless of lock state) — don't burn time on it.- Console's plain-text search bar does NOT accept typed boolean predicates — filter by eye or not at all.
5. When to stop fixing your app and file Apple Feedback
Escalation heuristic that saved the investigation: when (a) ≥2 well-reasoned app-side fixes (clean pack rebuild, full app reinstall) fail with byte-identical failure signatures, and (b) a cross-OS A/B test isolates the OS version (identical pack + device model: iOS 26.6 failed every attempt, 26.5.2 succeeded cleanly) — conclude platform bug. File Apple Feedback with the full evidence chain (error strings, A/B matrix, Console captures) and/or a Developer Forums post; stop guessing app-side.
6. CLI quirk: the upload wait-flag never exits
altool --upload-asset-pack's wait option can poll past terminal state without
exiting. Poll --asset-pack-status separately and kill the wait process once the
terminal state is confirmed.
7. ba-serve local dry-runs are not free
xcrun ba-serve (self-hosted local dry-run) requires its own build-variant setup —
materially more work than runbooks imply. Budget it as a real investigation task in
the host project before promising a "quick local test" of pack delivery.
04 同仓库更多 Skills
second-wind
Orchestrate long unattended Claude Code runs across multiple repos with the wind CLI. Use when the user wants to run Claude Code overnight, resume sessions after the 5-hour usage limit, run Claude in many repos at once via tmux, or says "set up second wind", "wind up", "overnight run", "resume after limit", "usage limit orchestrator".
structured-output-adaptive-thinking-token-bloat
A Claude API call using output_config.format (json_schema structured outputs) with thinking:{type:"adaptive"} burns most of its output budget on thinking (observed ~21k of ~24k tokens) without improving JSON quality, making the run up to ~5x more expensive than the same call with thinking omitted — use when calling the Messages API with a json_schema output format for extraction, ranking, classification, or summarization tasks, or when a schema-constrained call's usage.output_tokens is far larger than the returned JSON.
review
Apply a code-review lens to a diff, branch, or file — find correctness bugs, security issues, and quality problems, ranked by severity, with a concrete fix for each. Use when the user says "review this", "review my diff", "check this PR", "look this over before I commit", or "is this safe". Skip when the user wants the code written, not reviewed.
refresh
Rebuild the prompt-craft command registry and usage profile, and print a summary of installed commands and your personalized recommendations. Use when you install or update plugins and want the advisor to pick them up immediately, or say "refresh prompt-craft", "rebuild the command registry".
refactor
Apply a refactoring lens — restructure code without changing behavior, guarded by tests that pass before and after. Use when the user says "refactor", "clean this up", "extract this", "simplify this", "reduce duplication", or "make this readable" on existing working code. Skip when the change is meant to alter behavior (that's a feature or a fix).
plan
Decompose a non-trivial task into an explicit, tracked plan with goals and per-step acceptance criteria before writing code. Use when the user says "plan this", "break this down", "what's the plan", or asks for a multi-step feature/refactor that touches more than one file. Pairs with plan mode. Skip for single-edit changes that need no decomposition.