React Native Code One Task Group

This documentation was generated with the assistance of AI. Please report any inaccuracies.

Implement one bucket of React Native tasks from a plan’s task group end to end: capture a single pre-change quality baseline, run iru-react-native-code-one-task once per task (in parallel when allowed), then validate the entire bucket once — headers, TSDoc, Jest, an 80% unit-only coverage gate, type-check, native stages, and a lock-guarded device e2e run.

Purpose

iru-react-native-code-one-task-group is built on the same TypeScript gate skills as iru-typescript-code-one-task-group, each run through an iru-gate-runner agent so no raw report lands in its context, and adds four React Native-specific concerns. Native code has its own gates and the toolchain decides who owns it: for metro apps android/ and ios/ are committed source, so Step 3.7-3.8 build the side the bucket touched; for expo apps they are generated, so an edit there is a defect and Step 3.9 validates config-plugin changes by running expo prebuild in a throwaway copy. The coverage gate counts Jest unit tests only, since e2e flows and native tests produce no coverage of the measured TypeScript. tsc --noEmit is its own gate, because Jest presets strip types without checking them. The e2e stage needs a device, a build, and Metro, so it runs last and only while holding the repository-wide .integration-test.lock (acquired in Step 3.10a, released in 3.10b on every exit path; stale after 60 minutes, wait timeout around 45 minutes).

Step 1 detects the profile once and captures a baseline (plus an Android Lint baseline when metro tasks touch android/). Step 2 implements tasks, but tasks that edit the same navigator/layout file or native registration point run sequentially even in a parallelizable group. Step 3 runs license headers, TSDoc, scoped tests, coverage, the full suite, quality diffed against baseline, the Android, iOS, and Expo native stages as applicable, then e2e. Steps 4-5 finalize the plan notes and report; any unverified stage is named with its reason rather than read as a pass.

When the bucket touched UI, the device run of Step 3.10 doubles as the UI run (Maestro gallery capture flows under .maestro/ui/ and flow captures, or Detox, written into .ui-screenshots/), and a new Step 3.11 checks the evidence: every added/changed screen and component must have a capture for each declared state × .ui-quality.yml direction × theme on the platform that ran, and eslint-plugin-react-native-a11y findings from the lint gate are applied at the configured accessibility level (enforce fails the group, report lists them). No device makes the captures unverified, never passed. See Building user interfaces. With accessibility: none in .ui-quality.yml (opted out entirely), the accessibility rules and checks are neither applied nor required and are reported as "accessibility: opted out" in the group report, never unverified or failed; off still applies the rules but skips the automated checks. When .ui-quality.yml is missing in a repository that already has UI code, it doesn’t assume enforce: before the first UI task it asks once, pre-selected from iru-explore’s `Accessibility: line, and writes the file.

Purpose group

TypeScript Development Tooling

Plugin

Shipped by the iru-react-native plugin of the ai-catalog marketplace (install with /plugin install iru-react-native@ai-catalog; see Plugins).

Invocation

/iru-react-native-code-one-task-group <bucket text>

Inputs

Argument Required Description Default

Argument	Required	Description	Default
Bucket text	Yes	Every `react-native`-tagged task/sub-task in the bucket with its own text and checkbox lines, plus the group’s `Parallelizable` verdict and relevant "Current code state" context.	None — invoked with a specific bucket’s text by `iru-code-one-task-group` or `iru-typescript-code-one-task-group`.

Bucket text

Yes

Every react-native-tagged task/sub-task in the bucket with its own text and checkbox lines, plus the group’s Parallelizable verdict and relevant "Current code state" context.

None — invoked with a specific bucket’s text by iru-code-one-task-group or iru-typescript-code-one-task-group.

Outputs

Source, native, and config files plus tests per task, via iru-react-native-code-one-task.
License headers and TSDoc kept current for every file the bucket touched, checked once for the whole bucket.
Unit-only coverage of at least 80% line and branch for changed/new TS/TSX files, with any device-only file recorded as an accepted exception.
A type-check (tsc --noEmit), lint and format result compared to the single pre-change baseline.
Android, iOS, and Expo native-config stage results where applicable, each passed, failed-and-fixed, skipped as not applicable, or unverified with a reason.
An e2e (Maestro/Detox, optionally XCTest) result run under the shared lock, or unverified with the cause.
For UI buckets (Step 3.11): captures found vs expected per screen/component, malformed names, missing capture infrastructure (UI gallery, .maestro/ui/ flows), and react-native-a11y/* violations (count and rule ids) at the .ui-quality.yml level, or "UI captures unverified" with the reason.
implementation_plan.md with each task’s checkbox finalized with unit-only coverage, native stages, and e2e results reported separately; blocked tasks stay unchecked with their blocker note.
A report to the caller covering profile, per-task details, coverage exceptions, type-check results, native stages, e2e result, lock behavior, and which tasks need a new native build or are not OTA-compatible.

Execution flow

flowchart TD A[Start /iru-react-native-code-one-task-group] --> B["Step 1: detect profile once;\ncapture pre-change quality baseline\n(iru-typescript-code-quality, via iru-gate-runner)"] B --> C{Parallelizable and no shared\nnavigator or registration file?} C -- yes --> D["Step 2: run react-native-code-one-task\nfor every task concurrently"] C -- no --> E["Step 2: run tasks one at a time\n(or sequentially for conflicting ones)"] D --> F["Confirm/backfill each task's box;\nnotify user"] E --> F F --> G{expo toolchain and\nandroid/ios in diff?} G -- yes --> H["Defect: move change into\napp config, plugin or Expo module"] H --> I G -- no --> I["Step 3.1-3.6: license headers, TSDoc,\nscoped tests, 80% unit coverage,\nfull suite, lint + tsc + Prettier vs baseline"] I -- failure traced to a task --> J["Fix owning task;\nre-run affected checks"] J --> I I -- clean --> K{Native stages needed?} K -- metro android/ --> L["Step 3.7: gradlew lint, JVM tests,\nassembleDebug from android/"] K -- metro ios/ --> M["Step 3.8: pod install +\nxcodebuild build (macOS/Xcode only)"] K -- expo config/plugins --> N["Step 3.9: expo config introspect +\nprebuild in throwaway copy"] K -- none --> O L --> O M --> O N --> O{e2e wired and device available?} O -- no --> P["Report e2e unverified or not wired"] O -- yes --> Q["Step 3.10a: acquire\n.integration-test.lock (wait, stale break, timeout)"] Q --> R["Step 3.10: ensure app current, run\nMaestro/Detox via iru-typescript-test e2e"] R --> S["Step 3.10b: release lock\non every exit path"] S --> UE{"Bucket touched UI?"} P --> UE UE -- yes --> UV["Step 3.11: check .ui-screenshots/ against\nstates x directions x themes;\napply a11y level"] UV --> T["Step 4: finalize each task's note\nwith unit, native and e2e outcomes"] UE -- no --> T T --> U["Step 5: report the bucket's\noutcome to the caller"]

Dependencies

Invokes

React Native Code One Task — Step 2, once per task in the bucket (concurrently or sequentially per the Parallelizable verdict, with conflicting registration-point tasks forced sequential), via the iru-isolated-skill-executor agent.

Each of the following is invoked via the iru-gate-runner agent rather than directly:

TypeScript Code Quality — Step 1 (pre-change baseline) and Step 3.6 (ESLint, tsc --noEmit, Prettier diffed against it).
Check License — Step 3.1, license headers for every file the bucket touched.
TypeScript TSDoc — Step 3.2, TSDoc for exported TS/TSX symbols.
TypeScript Test — Step 3.3 (scoped Jest tests), Step 3.5 (full unit suite), and Step 3.10 (e2e selector with lock: held).
TypeScript Coverage — Step 3.4, unit-only coverage gate.
Android Code Quality and Android Test — Steps 1 and 3.7, for a metro app’s android/ side, only when they can target the android/ Gradle root; otherwise plain ./gradlew commands run through the same agent.

Invoked by

Code One Task Group — Step 2, bucket for key react-native.
TypeScript Code One Task Group — Step 1, hands the whole bucket over when the detected flavor is React Native and this skill is installed, and relays its report.

Related agents

iru-isolated-skill-executor — spawned once per task (Step 2), many at once when parallelizable.
iru-gate-runner — spawned for the baseline (Step 1) and every validation, native-build, Expo-check, and e2e stage (Step 3), keeping reports out of this skill’s context.

Source

SKILL.md on GitHub — the file this page was generated from.