cd1e13c1e2
browser tests run only by hand: the suites that launch Chrome / Chromium move to tests/browser/*.browser.ts (bun test, ci, ci:quick and the Forgejo workflows never pick them up; bun run test:browser runs them); the desktop boot check leaves bun run ci (it stays in desktop-release.ts mac at release time); CI=full is gone: Forgejo's ci runs the Rust steps when the push changed a Rust input (STOREBOUND_CI_BASE=HEAD~1), no Playwright install; CONTRIBUTING
2026-10-15 01:25:17 +08:00
4c7ff89950
docs/pitfalls.md: rows 64-67 (the skill had them)
2026-10-15 01:21:13 +08:00
6873f2cc68
release candidates: draft assets are read by their attachment uuid (the tag URL is 404 for drafts: the record job could not merge SHA256SUMS); the installers are scanned with the private terms in their build jobs, so the record job needs no apt tools (it spent 5 minutes installing them)
2026-10-15 01:18:55 +08:00
c3b1e8c370
release candidates: every push to main runs the full CI, the acceptance matrix and the Linux / Windows installer builds in parallel (ci.yml) and records the green result per commit as the draft rc-<sha> (green.json last); a version-only bump commit (proven file by file with the version writers) reuses its parent's CI and matrix and rebuilds only the installers; a v* tag only promotes the candidate (release.yml), fails fast on a red or missing commit; acceptance runs carry a CLI key; tools/release.ts does the whole release on the Mac (bump, CHANGELOG cut, fast gate, push, tag, Mac build + notarize in parallel with Forgejo, cached macOS acceptance, add, verify, publish with one upload of the Mac files, /Applications update, npm commands) and --rehearsal; DMG notarization runs while the app and updater bundle are checked; rehearsal channel on the VDS; docs
2026-10-15 00:42:50 +08:00
89d94d6993
console Chrome never outlives its run: exit kills it, SIGINT / SIGTERM / SIGHUP close it gracefully, a detached watchdog ends it when the parent dies (SIGKILL included; a reused pid is never killed); every start ends headless orphans on storebound profiles and names orphaned windows; headless runs on macOS prefer Chromium / Chrome for Testing over Google Chrome.app (an orphan kept the owner's Chrome from opening from the Dock); console fill closes the window on errors; pitfall 68; tests
2026-10-15 00:28:32 +08:00
b350781ab5
acceptance: the app fixture carries the owner's answers (unanswered: [], its own name and copyright; template answers are refused before store writes); macOS results of 0.8.4 (123/123 on Node 20, Node 22, Bun), docs/acceptance.md regenerated
2026-10-14 23:56:57 +08:00
357319a775
--json prints only JSON on stdout (withJsonOutput around every command: logs, tables and prompts to stderr; printJson everywhere; test runs every --json command against the fake stores and parses stdout); prices ppp compares like with like: par = the store's own US-equivalent price (Play convertRegionPrices, Apple equalizations; offline base x fx x (1 + VAT) rounded up) scaled by the PPP price level; defaults only lower, only price levels <= 0.5, World Bank high-income at par, CN at par on Apple and skipped on Play, never above par (--only-decreases, --max-ratio, --min-ratio, --max-level, --include-high-income; --floor / --ceiling kept), top-10 revenue market drops > 20 % warned (sales reports, else a static list), unchanged regions out of the sheet; price changes of existing products only with --allow-price-changes, TTY or not (a terminal yes covered them), then their own question listing each old -> new, held section in dry runs (play products, prices import, play / apple subscriptions); pitfalls 61-63; tests
2026-10-14 17:36:00 +08:00
7219ded3e8
template compliance answers never reach a store: init writes unanswered (dataSafety, appPrivacy, reviewNotes, ageRating, accounts, reviewAccess), older configs are recognised by the template values; check / validate / scan block, play data-safety (also --dry), console plan / fill, apple review-details and appinfo refuse. onboard --declarations replaces them on a live app: name and copyright from the stores (a copyright with a private term is left out), the rows the code proves (scan proposal, sign-up methods from auth code and login SDKs), the rest as one owner question list. play data-safety --dry prints the answers that change (old -> new, account and deletion first) against the last uploaded CSV; pitfall 60; tests
2026-10-14 11:36:30 +08:00
34c2fa22e5
site: first-party, cookieless visit stats (self-hosted Umami at storebound.dev/s/, Do Not Track respected) with events for copy buttons, desktop downloads per OS, the teaser (play, 50 %, complete) and outbound docs / git / npm links; a privacy note in the footer; docs/site-analytics.md and tools/vds/ configs to rebuild the server side; scrub allows a page's public stats website id
2026-10-14 11:11:24 +08:00
a8c7f77d87
acceptance: macOS results of 0.8.3 from 1c44095 (123/123 on Node 20, Node 22, Bun), docs/acceptance.md regenerated
2026-10-14 08:51:14 +08:00
9d6f741092
acceptance.yml: the release matrix on demand (Linux + Windows x node20/node22/bun, then the report); macOS results of 0.8.3 (123/123 on Node 20, Node 22, Bun) in docs/acceptance/, docs/acceptance.md generated
2026-10-14 08:38:12 +08:00
ae681e3ecd
release: privacy scan of signed macOS artifacts leaves out Apple code signatures only (copies with codesign --remove-signature on every Mach-O, the DMG's koly signature blob zeroed); any other hit still fails; tests and docs
2026-10-14 08:21:35 +08:00
00be838b76
store art from the live stores: storebound art pull (App Store screenshot sets and previews per locale and display type from imageAsset.templateUrl at the original size, the App Store icon from the newest build or the public artwork; per Play language icon, feature graphic, phone / 7" / 10" / TV / Wear screenshots as the =s0 originals checked against Play's sha256) into store/renders with pulled.json (sha256, store, URL host, store checksum); upsert (a re-run changes nothing, differing local files kept unless --force, the upload's file of a slot compared: Play .jpg, App Store .png, the App Store icon as a picture); onboard runs it; apple screenshots / previews and play images see unchanged pulled files as the store's own; check: render sources are notes with pulled art; local renders check: pulled files pass, store-empty sets not demanded; art live / thumb / replace; desktop Store art tab: live thumbnails per store, language and set (same / differs / not pulled), Pull from stores, store-only apps read-only, Replace with a file (image_pick, art.replace) in the lightbox and the art editor; boot check store art case; tests/art-pull.test.ts; console parity 76 rows (57 complete, 19 with the reason); store-screenshots skill (pull first for live apps), pitfall 59
2026-10-14 05:25:37 +08:00
7ac028421a
store-only apps edit their listing on the store (storebound listing live / set: old -> new per field with the checks, a PATCH of only the edited fields; serve listing.set is the one store-only write, its token bound to the store ids; policy.rs STORE_ONLY_WRITES, write_with_confirm for store targets); screenshot captions per language and scene edited as a text edit of the shots copy JSON (JSON files keep their layout); Play content rating, target audience and App content answers in App info; docs/console-parity.md after the editors (73 rows: 55 complete, 18 with the reason) and tests/console-parity.test.ts keeping it honest
2026-10-14 00:54:56 +08:00
3cea2af4e1
desktop editors: edit in place what App Store Connect and Play Console let the owner edit (listings per language for both stores with live counters and the check / scan rules, store details, App info, age rating, App Review, Play store settings, Data safety and App Privacy answers, promotion blocks, catalog names / prices / offers, TestFlight information, scene order); storebound edit (project.edit) saves into the config by path keeping comments and formatting, the listing files and the review contact, dry run with the diff and the checks first; Apply shows each store's dry run in the window; unsaved edits survive tab switches with undo / redo; tests/edit.test.ts, apps/desktop/tests/edits.test.ts, boot check editor case
2026-10-14 00:37:08 +08:00
e59f439101
docs/console-parity.md: audit of every App Store Connect and Play Console field storebound reads or writes against the desktop (view, edit in place, store write), with the gaps
2026-10-14 00:09:54 +08:00
7b05e4b396
auto-read after the rebase: results through usable / failureText, issuesError notice kept, scheduler 'accounts' (no credentials word in the webview), fixtures and generated tables for health_issue
2026-10-13 23:41:36 +08:00
0fb4ce0001
desktop reads everything by itself: one background read scheduler (4 per store, starts spaced per account, App Store Connect X-Rate-Limit slows it below 20 % and pauses at 100 left, backoff with jitter on 429 / RESOURCE_EXHAUSTED, visible app first, one read per key), per-method TTLs over the saved reads, Settings auto-read (on) and interval (15 min, paused while hidden), Refresh all, a refresh icon and 'read 3 min ago' / 'couldn't refresh' per card, writes invalidate, status line with progress and quota; quota reaches the window through ToolResult.quota (cli/lib/quota.ts); Health top issues open with versions, sample stack (health --issue / --signature, read health_issue), Open in Play Console, Copy for AI and likely-cause hints (cli/lib/crash-hints.ts); tests/scheduler.test.ts, tests/quota-crash-hints.test.ts, boot check auto-read case
2026-10-13 23:37:39 +08:00
19d34286d5
Feature graphic fits the phone: shots.feature.fit 'contain' (default) sizes the device so its rotated box stays inside 1024x500 with a 24 px margin in both templates, 'bleed' keeps the crop, scale 0.5-1.5, config validated; shots render records measured device boxes (report.json devices) and shots check fails a feature graphic whose device leaves the canvas (warns for a screenshot's front device off a side or the top); tests/feature-fit.test.ts (rotated-box geometry, both templates rendered at 1024x500, bleeding fixture) (pitfall 55)
2026-10-13 09:50:15 +08:00
0e4f579899
apple build exports each build into build/ios/export/<version>-<build>/ (cleaned first), runs xcodebuild with Apple's tool folders first on PATH (Homebrew rsync 3.x broke Xcode's IPA step: exportArchive Copy failed), shows the IDEDistribution log's real error lines and exits non-zero on any export failure; apple upload fails with exit 1 when the IPA's build number is already on App Store Connect (stale IPA after a failed build), --allow-existing re-checks, --build N picks that build's IPA, release and beta flows use both; doctor flags a non-Apple rsync first on PATH; desktop publish calls rsync by absolute path without --chmod (publishSteps); site download notice per OS (macOS signed and notarized, Windows unsigned for now, Linux unsigned); game-report credential tests hermetic against a machine default profile (pitfalls 53, 54)
2026-10-13 09:45:48 +08:00
2590b2e828
Releases on git.crt.fyi from a workflow: a v* tag push runs .forgejo/workflows/release.yml (tag = package.json version and every version copy, the full CI, Linux AppImage + deb and Windows NSIS on the VDS runners, privacy scan with the private terms, one Forgejo release with short CHANGELOG-summary notes, the installers and SHA256SUMS, older release entries deleted, tags kept); workflow_dispatch dry run makes a draft test-<sha> and deletes it; desktop.yml folded in, ci.yml and release.yml share .forgejo/actions/linux-setup and ci-run; tools/forgejo-release.ts (notes, upsert, prune, delete; uploads over 100 MB go to the origin because Cloudflare refuses them); desktop-release.ts upload-forgejo adds the notarized DMG and updater bundle from the Mac; docs/desktop-release.md release order
2026-10-13 07:44:18 +08:00
e5f148e534
Onboarding live apps without owner questions for what the stores and the code prove: storebound onboard imports the live products and subscriptions with USD / base prices (catalog diff clean), contact values, the App Review contact and demo account (credentials.json, masked), sign-in for review from Apple's review detail and auth code, and the account deletion page from the app's site; App Store keywords count characters (live fields up to 265 bytes accepted); validate accepts the 6.5-inch iPhone set and Play's default-language graphics fallback; Play Reporting API off gives the project's enable link and apps list falls back to local apps; accounts default <profile>; store-only apps (apps add --store --app-id) shown read-only in the desktop (pitfalls 43-52)
2026-10-13 07:25:01 +08:00
d2ce185441
One product per machine: the CLI and the desktop app share one version, the store accounts and the app folder list
...
- one version: package.json's version is the desktop app's too (desktopVersion removed); tools/desktop-version.ts
keeps the plugin manifests and the app's copies equal, desktop-release publish refuses another version, release
order app first then npm publish; the unpublished desktop 0.1.0 candidate is noted in the docs and CHANGELOG
- storebound desktop installs (per user, no admin: macOS DMG -> /Applications or ~/Applications, Windows NSIS /S,
Linux AppImage + .desktop with the storebound:// handler), updates an older app, notes a newer one (downgrade only
with --force), asks once ([Y/n], --yes, exit 3 without a terminal), opens on the folder; --dry / --json / --download;
the install steps are pure plans (cli/lib/desktop-install.ts); the version's own manifest dl/<version>/latest.json
- credentials: one store, the CLI's credential profiles in ~/.storebound/credentials; the desktop writes and reads
them (profiles.rs, files 0600), moves old keychain accounts into profiles on first start, shows profiles the CLI
made; precedence env > named profile > project file > default profile; config `account`, "account" in the
project's credentials.json; storebound accounts list | import (dedup by key id / service account e-mail, key files
copied 0600, project file keeps keystore and review contact, backup, --dry, never prints values) | use
- app folders: one list ~/.storebound/projects.json shared by the desktop (registry.rs), storebound desktop,
apps folders | add | remove, ws init and ws without a workspace file
- desktop: notice when a folder pins another storebound version; storebound-desktop --version
- tests: version drift, desktop decisions and per-OS plans, credential precedence and import, registry (TS + Rust),
keychain migration and profile round trip (Rust), boot check with old keychain state and the new shared state
2026-10-13 06:17:37 +08:00
bba46b5c7f
apple testflight distribute: a build's groups read per group (GET /v1/betaGroups/{id}/relationships/builds) instead of GET /v1/betaGroups?filter[builds]=, which App Store Connect refuses with 400 although the spec declares it; internal all-builds groups reported as already available and never sent the build; a re-run adds nothing; configured groups missing on App Store Connect point at groups sync, unknown ones list what exists; APPLE_LIVE_REJECTED (refused by storebound api too), tests/asc-query-guard.test.ts checks every App Store Connect GET with a query in cli/ against the spec's declared parameters, and the fake App Store Connect answers 400 to undeclared or live-refused parameters (pitfall 42)
2026-10-13 06:03:16 +08:00
f741c15829
console: Play App content card Saves are the agent's (declaration-save; Send changes for review, Submit, Publish, Restart review stay forbidden), one owner batch per plan with the IARC Terms tick before the run, console plan play-app-content --all, New App fields typed not set, play.appId from the Create app redirect (console verify --url --write) and storebound play app-id across app list pages
2026-10-13 05:44:29 +08:00
1366e7e231
scan, privacy answers and credentials from a second game's first release (lessons 35-37): review-prompt needs rating context, not a bare "rate"; account signals skip import lines and module paths, config-only accounts say "from config"; accounts.inApp without deletion code stays a medium 5.1.1(v) finding; privacy-manifest --propose builds NSPrivacyCollectedDataTypes from appPrivacy.types and check warns on an empty list; appPrivacy.types is authoritative in the App Privacy sheet (unlisted = not collected), ad / support / AR types default to not collected without tracking or ad SDKs; Play advertising ID from SDKs, the AD_ID permission, app-ads.txt or the flag, never from a Device ID alone; scan.ignore[].evidence and warnings for entries that match nothing; empty credentials template, placeholder values count as missing (check, doctor)
2026-10-13 05:39:09 +08:00
86cd79dd45
desktop-release add: evidence-linux / evidence-windows files (run, commit, sizes, hashes, scan) and a check that every artifact of a version comes from one commit (--commit against the folder's other evidence); docs: check the downloads against the run before add
2026-10-13 05:09:17 +08:00
36a8bf9625
CLI runs on Node 20+: npx storebound (bunx keeps working)
...
- cli/lib/rt.ts: the runtime layer on node: modules (spawn / spawnSync with Bun's defaults, exit codes and
"Executable not found", which with PATHEXT, sleep, serve, Glob, hasher, gunzip, stdin, SQLite, packageRoot,
config imports with a type-stripping fallback for Node 20); every Bun.* use in cli/ goes through it
- tools/gen-embedded.ts -> cli/lib/embedded.ts: templates/, cli/shots/*.html and Play's Data safety form embedded
(replaces Bun text imports; the bundle and the compiled sidecar carry them), checked fresh by bun run ci
- package.json: bin dist/storebound.mjs (Node ESM bundle, tools/build-node.ts, prepack), engines node >=20
- bun run ci: node bundle + node smoke (real node: --version, help, check same as Bun, scan, release plan --help,
init, MCP handshake + tool call); Forgejo CI sets up Node 20; tests/node-compat.test.ts guards cli/
- install writes npx MCP commands when run with Node; bin/storebound falls back to Node
- docs: npx storebound init (Node 20+) or bunx storebound init
2026-10-13 04:03:28 +08:00
710c3a53d0
README, docs and site: one story (the whole job, no avoidable rejections, the desktop app, your AI, install)
...
README is a short front page (about 120 lines) with one real visual per section; every former section moved
unchanged into docs/*.md with a docs/README.md index (journey, rejections, getting-started, commands, growth,
studios, store-art, listing-languages, api, skills, mcp, safety, faq, project). The site tells the same story on one
page: the teaser in the hero, real CLI output as text, the desktop app as a section with light/dark captures of eight
screens and the per-system downloads (desktop/latest.json); /desktop/ redirects to /#download. New capture tools:
tools/site-desktop-shots.ts (fixture-mode desktop frames) and tools/site-term-shots.ts (real CLI output as text and
WebP). tests/docs-links.test.ts checks README, docs and site links resolve.
2026-10-13 03:29:59 +08:00
12b01f065c
desktop-release: architecture guard before signing (tauri build --no-sign, guard, then tauri bundle signs / notarizes): every Mach-O in the app must contain arm64, nothing Intel-only, Info.plist must not force Intel (LSRequiresNativeExecution false, LSArchitecturePriority x86_64 first); guard repeated on the signed app, the DMG and the updater bundle, recorded in evidence.json
2026-10-13 02:32:09 +08:00
9ba8471002
ci speed: desktop boot check runs its cases in parallel (one browser, killed at the end; a page per case; condition waits instead of fixed sleeps: element, text or a quiet window of host calls and DOM); shots capture / render close the browser with a 3 s cap, then kill it; console engine waits injectable for tests (WAITS / engineWaits: poll rounds, a missing step ends once its loaded page stays quiet, team switch settle, field-miss grace), fillPage looks up each data type's dialog once per round and skips dialog sections when no dialog is open; browser tests run 4 at a time per file (testScope().concurrent with a scope per test, --max-concurrency in ci), 20 ms rounds; capture-web counts boots per test and runs its browser tests concurrently; console-session / -persist Chrome tests concurrent; recorded test timings smoothed (mergeTimings) so one loaded run does not skew the shard split
2026-10-13 02:22:25 +08:00
f2d7062bcb
docs desktop-release: the Linux job starts the AppImage and the deb under Xvfb
2026-10-13 02:09:58 +08:00
3952716ef8
desktop-release: never run the x86_64 slice on the build Mac (Rosetta makes macOS warn the owner that the app will not open in the next macOS); every Mach-O in the bundle checked with lipo (arm64 + x86_64), self-test and sidecar version run natively only; pitfall 34
2026-10-13 02:01:35 +08:00
2b101f0fd8
desktop: Releases > Send for review (App Store submission state, builds by number, Submit / Resubmit / in-app purchases; Play versionCodes by number, Send to production, managed publishing, console plans for countries and Send changes for review), plan panel before every owner dialog with streamed progress; App info console steps for every console plan form with verify; Catalog price-change owner actions and the declined state; Data safety deletion answer; TestFlight create missing groups; Android SDK in Dependencies; upload progress in job toasts; boot check send-for-review cases on D1 and D3 state
2026-10-12 12:08:43 +08:00
33eb24a3dc
apple submit --build <n> (first submission, the resubmit steps; plan table with --json under serve); desktop owner actions to send for review: owner.apple_submit, owner.apple_resubmit, owner.apple_iap_submit, owner.play_production, and the price gate owner actions (owner.prices_play_products / _play_subscriptions / _apple_subscriptions; owner.prices_import passes --allow-price-changes after the dialog); OWNER_ALLOWED_ARGS ties each forbidden flag to its one action, dry runs never carry owner flags, MCP still has no submit or production release; submit dialog text in the host; system.tools androidSdk, project.inspect playRelease, declarations dataDeletion
2026-10-12 12:08:07 +08:00
0b1451800c
apple iap-submit: DEVELOPER_REJECTED IAP / subscription / group versions (left by a canceled submission) are submitted as they are, plan says 'add (developer rejected, re-submitted as is)', --new-version only for APPROVED / REJECTED, empty draft reused; fake ASC models cancel and re-add. play build / check: warn before Gradle when defaultConfig does not read -PversionCode / -PversionName, with the exact Groovy / Kotlin line (built with AGP 8.13); android.injected.version.code does not override bundleRelease; post-build versionCode mismatch is a hard stop. play-production-release plan: 'Do you want to restart your review?' dialog, Restart review is never-press. Pitfalls 31-33
2026-10-12 11:32:05 +08:00
41d2ffcc24
apple testflight groups: bare lists the groups and what the config would create (read-only); groups sync (or --yes) creates; flows: bare = read, sync = write; MCP tool runs groups sync
2026-10-12 10:48:59 +08:00
1724296e59
apple resubmit: REJECTED review submission items are marked resolved before the submit (PATCH reviewSubmissionItems resolved: true; the version item by default, IAP / subscription items with --resolve-iaps), plan table 'mark resolved' rows; on 409 'not ready yet' the items are re-read and a REJECTED item out of scope stops the run at once naming it and how to resolve it, backoff only when none is left; an empty READY_FOR_REVIEW draft is never canceled (409 'not in cancellable state'): left as is, no cancel command, apple status labels it harmless and treats the cancel 409 as such; fake ASC blocks submit on REJECTED items and refuses to cancel empty drafts. Pitfall 30
2026-10-12 10:45:05 +08:00
0da5ee9d01
apple resubmit after a rejection: iap-submit reuses the open review submission (UNRESOLVED_ISSUES gets only missing items and is resubmitted, a draft is reused, one waiting for or in review means nothing is sent), never a second draft; 409 ITEM_PART_OF_ANOTHER_SUBMISSION names the other submission instead of the list-lag message, a draft the run created is canceled and the named one used; submit retried with backoff and heartbeat while Apple says not ready yet; new apple resubmit --build <n> (build by number, plan table, --yes); apple status lists UNRESOLVED_ISSUES items and the resubmit command. Pitfall 30
2026-10-12 10:34:12 +08:00
c8213be2b6
ITMS-91064: NSPrivacyTracking and NSPrivacyTrackingDomains must agree in every privacy manifest (true needs domains, domains need true); check, scan and privacy-manifest fail on either mismatch with the exact ITMS text; App Privacy tracking no longer demands the app manifest's true: SDK manifests that declare tracking (Pods, frameworks, SwiftPM, Carthage, plugins, the built IPA) are info, known tracking SDKs info, else a warning; privacy-manifest --propose writes true only with appPrivacy.trackingDomains; apple upload checks every manifest in the IPA (XML or binary) before any network call, apple build after the export. Pitfall 29
2026-10-12 10:27:21 +08:00
b7bfe89151
Play console steps without an API as agent plans: play-store-settings (App category; play listing cannot set it), play-app-content (every card in order, one owner batch) and its cards (privacy policy, ads, sign in details, advertising ID, government, financial, health), play-production-countries and play-production-release; IARC plan with the Terms of Use agreement, follow-ups (console.contentRating.details, products[].randomRewards, social) and the Summary's final Save; config play.category / countries / managedPublishing, console.appContent; check, validate, play status and play release reminders; inPageLoop helper and material checkbox clicks; App Privacy published signal; first IAPs via apple iap-submit --with-version. Pitfalls 27, 28
2026-10-12 10:16:34 +08:00
e01a27d7b7
price changes of existing products need --allow-price-changes (play products, play/apple subscriptions, prices import): --yes and flow run --apply --yes confirm creates, first prices and other updates only; without the flag the run writes the rest, names each held change and exits 3 (declined), dry runs mark them owner, flows refuse the flag; lib/price-gate.ts, tests/price-gate.test.ts. CLI wording: console fill leaves what it cannot open or fill to the calling agent's browser tool (console plan / verify) instead of 'yourself'; other 'by hand' output points at the config; do-it-dont-ask test scans cli/ output
2026-10-12 09:38:57 +08:00
b5b8f2da9d
play data-safety: PSL_SUPPORT_DATA_DELETION_BY_USER as Play's single choice (partialDeletion true | false | 'auto90'), drop the removed PSL_DATA_COLLECTION_USER_REQUEST_DELETE (deletionRequest deprecated, check notes it); ship Google's empty form (cli/data/play-data-safety-form.csv, exported 2026-10-12), take every Answer requirement from it and report pairs not in it (formPairIssues) in check, declarations and play data-safety. Pitfall 21
2026-10-12 09:34:12 +08:00
dfc28c15f8
agent rules: first rule "Do it, don't ask" (agents-core and every adapter, README, store-pitfalls lead, first-release, declarations and console skills); the agent runs the real command after the dry run, the owner presses only final public or irreversible buttons and answers compliance facts in one batch; handoff text removed from skills; tests/do-it-dont-ask.test.ts. Pitfall 26
2026-10-12 09:28:45 +08:00
fc524cd2d4
play build: find the Android SDK like Android Studio (ANDROID_HOME, ANDROID_SDK_ROOT, sdk.dir, default folder), pass it in the env, stop before Gradle when missing; doctor shows it. apple upload: size, per-part lines, 20 s heartbeat (api, altool, processing wait), per-part and per-call timeouts, --timeout exits 7 with the resume hint. Pitfall 25
2026-10-12 08:59:45 +08:00
366362d62b
flow run: a declined confirmation stops the flow; --apply --yes confirms write steps
...
confirm() prints 'declined: <reason>' and exits 3 whenever the answer is no or nobody can answer (not a
terminal without --yes, serve without the owner's dialog), so no command reports success after declining.
flow run stops at such a step with 'step N declined: <reason>' (status declined), never ok. flow run
--apply --yes passes --yes to write steps only; --yes without --apply is refused. Pitfall 24.
2026-10-12 08:51:58 +08:00
5fbb9e63e3
build.prebuild and a Capacitor web-asset staleness guard for apple build / play build
...
apple build and play build run build.prebuild (or apple.prebuild / play.prebuild) first, once per flow run;
--no-prebuild skips it, --dry only prints it. A Capacitor project's copied public/index.html must then be newer
than the last source commit and <webDir>/index.html, else the build stops (--allow-stale-web overrides). check
warns when stale without a prebuild; the build output and dry plan show the web bundle's sha256 and mtime.
The Capacitor project is also found in a monorepo (apps/native etc.); several ask for play.build.dir /
apple.build.*. Pitfall 23.
2026-10-12 08:51:38 +08:00
f694af80eb
pitfall 22: AAB proto manifest field numbers
2026-10-12 08:46:03 +08:00
720cb2a13a
play images: same languages as play listing (drafts skipped unless --include-drafts); a language with no Play listing (404) is skipped with an info line, never fatal
2026-10-12 08:45:43 +08:00
97ab76e9c8
play data-safety: write the conditional required rows (PSL_ACCOUNT_DELETION_URL from accounts.deletionUrl, PSL_SUPPORT_DATA_DELETION_BY_USER from dataSafety.partialDeletion, PSL_DATA_DELETION_URL); a rule table checks every conditional row in check, declarations and --dry before Play answers 400
2026-10-12 08:40:36 +08:00
