Android workmanager notifications

Schedule reliable background work, reminders, and notification delivery with WorkManager and Android execution limits.From its SKILL.md

Install

npx -y skills add krutikJain/android-agent-skills --skill android-workmanager-notifications

Assembled from the repository path, not quoted from the project. Check it against their README if it does not work.

One thing to look at

14 stars14 stars. Stars are a popularity signal and not a quality one, but at this level it is likely that nobody has read this closely except its author, and you would be relying on your own review.

5.6 KB, 970 tokens by cl100k_base, as published. Nobody here has run it

Use this skill when the request is about: android workmanager job, background reminder notification android, reliable retry workmanager.
Primary outcome: Schedule reliable background work, reminders, and notification delivery with WorkManager and Android execution limits.
Read references/patterns.md when you need the API-choice matrix for WorkManager vs foreground service vs alarms.
Read references/scenarios.md for unique-work, notification-permission, and reminder-delivery review paths.
Handoff skills when the scope expands:
android-permissions-activity-results
android-performance-observability

Choose the right execution API first: WorkManager for deferrable guaranteed work, foreground services for ongoing user-visible work, and exact alarms only when clock-time precision and policy requirements truly demand them.
Make work idempotent and uniquely addressable with explicit names, input data contracts, backoff policy, and cancellation behavior.
Apply constraints, expedited work, and retry semantics only where they match user value and platform limits.
Build notification delivery as part of the contract: channels, importance, tap action, runtime permission handling, and graceful behavior when notifications are denied.
Validate duplicate scheduling, reboot/process death, and degraded states before handing off performance or permission-specific follow-up work.

Keep background work idempotent and cancellation-aware.
Use unique work or unique periodic work when the feature represents one logical reminder or sync pipeline.
Treat expedited work as scarce and user-valuable; do not use it as a generic queue-jumping flag.
If a worker uses network constraints, ensure the app declares the permissions required for those constraints on modern Android versions.
Keep notification value intact even when the user declines POST_NOTIFICATIONS; the app still needs a visible recovery path.

Blocking the main thread with disk or network calls.
Expecting WorkManager to deliver exact wall-clock timing like an alarm API.
Re-enqueueing duplicate reminder work instead of using unique work policies.
Failing to create channels or handle notification permission decline paths.
Hiding critical foreground or exact-alarm policy assumptions in comments instead of code and docs.

Is WorkManager the right API for this workload?
Are unique work policies, constraints, retry, and cancellation explicit?
Do reminder and notification paths still make sense if notifications are denied?
Are expedited, foreground, or exact-alarm behaviors justified and policy-safe?

Scenario: Find background-work and notification hooks before proposing a scheduling change.
Command: rg -n "WorkManager|Worker|enqueueUnique|PeriodicWorkRequest|NotificationChannel|POST_NOTIFICATIONS" examples

Scenario: Review exact-alarm, expedited, or network-constrained work for platform-sensitive behavior.
Command: rg -n "setExpedited|setRequiredNetworkType|AlarmManager|ForegroundInfo" examples

Scenario: Disambiguate WorkManager requests from permission prompts and performance work.
Command: python3 scripts/eval_triggers.py --skill android-workmanager-notifications

The chosen execution API matches the workload's timing and visibility requirements.
Unique work, retry, cancellation, and constraint behavior are explicit.
Notification channel and permission behavior are handled as part of the feature contract.
Process-death, reboot, and duplicate-scheduling behavior have been considered.