timers-alarms-reminders

[FHLiang221]

What does this Ability do?

A combined interactive + background daemon ability that handles timers, alarms, and reminders through natural voice interaction. Users can set, list, cancel, and delete events. A background watcher fires events when due — timers get a spoken notification, alarms play a sound, and reminders speak the message.

Suggested Trigger Words

• set a timer
• set an alarm
• remind me
• wake me up
• timers
• alarms
• reminders
• schedule

Type

• New community Ability
• Improvement to existing Ability
• Bug fix
• Documentation update

External APIs

• No external APIs
• Uses external API(s):

Testing

• Tested in OpenHome Live Editor
• All exit paths tested (said "stop", "exit", etc.)
• Error scenarios tested (API down, bad input, etc.)

Checklist

• Files are in community/timers-alarms-reminders/
• main.py follows SDK pattern (extends MatchingCapability, has register_capability + call)
• README.md included with description, suggested triggers, and setup
• resume_normal_flow() called on every exit path
• No print() — using editor_logging_handler
• No hardcoded API keys — using placeholders
• No blocked imports (redis, connection_manager, user_config)
• No asyncio.sleep() or asyncio.create_task() — using session_tasks
• Error handling on all external calls
• Tested in OpenHome Live Editor

Anything else?

Includes a background.py background daemon that polls scheduled_events.json every 5 seconds and fires due events. Also writes upcoming_schedule.md so the Personality stays aware of what's scheduled. Includes alarm.mp3 for alarm sounds.

[github-actions]

🔀 Branch Merge Check

PR direction: add-timers-alarms-reminders → dev

✅ Passed — add-timers-alarms-reminders → dev is a valid merge direction

[github-actions]

✅ Community PR Path Check — Passed

All changed files are inside the community/ folder. Looks good!

[github-actions]

✅ Ability Validation Passed

📋 Validating: community/timers-alarms-reminders
  ✅ All checks passed!

[github-actions]

🔍 Lint Results

✅ __init__.py — Empty as expected

Files linted: community/timers-alarms-reminders/background.py community/timers-alarms-reminders/main.py

✅ Flake8 — Passed

✅ All checks passed!

[uzair401]

Hey @FHLiang221, great work on the structure here — the LLM-routed intent classification is the right approach and the background watcher pattern is solid. Found a few things that need to be fixed before this can be approved though:

1. The delete and list fast-paths in _process_request use hardcoded in lower substring matching before the LLM ever sees the input — a user saying "get rid of my alarm", "wipe everything", or "what have I got set" will silently fall through to the wrong action or no action. Remove both fast-paths entirely and let the LLM handle DELETE_ALL and LIST_EVENTS — it already does this correctly.

2. QUIT_WORDS and EXIT_WORDS are missing too many natural spoken variants to be reliable. QUIT_WORDS is missing: "stop", "cancel", "forget it", "nevermind" (one word — common STT output), "never mind that", "leave it". EXIT_WORDS is missing: "no thanks", "I'm all set", "that'll do", "that's all", "we're good", "all good". Long-term fix: replace _wants_to_quit and _wants_to_exit with a single LLM classifier — "Is the user trying to stop or exit? Return YES or NO."

3. The affirmatives set in _process_request is missing: "absolutely", "go ahead", "do it", "sounds good", "why not", "let's do it", "uh-huh", "please". Same fix — swap for an LLM classifier.

4. _is_bare_trigger uses rstrip("s") to handle plurals, which silently corrupts any word ending in "s" that isn't a plural. Replace with an explicit set: {"alarm", "alarms", "timer", "timers", "reminder", "reminders", "schedule"} mapped to their canonical form.

5. _build_system_prompt has no instruction for how QUESTION: responses should sound — that text goes directly into speak() and could come back with bullet points or markdown. Add to the prompt: "Questions must be plain spoken English — no bullet points, no markdown, no lists. One short question only, under 15 words."

6. _handle_list can produce 50–70 words in one unbroken sentence when 4–5 events are scheduled. Cap the spoken summary at 3 items and append "and X more" for the rest.

7. "Done. Your %s %s has been cancelled." — word order sounds unnatural on TTS. Change to "Done. Your %s %s is cancelled." — drops "has been", cleaner output.

8. "Quitting ability." appears three times in first_setup — robotic, leaks implementation detail. Change to "All done." consistently.

9. "All done. Handing you back." — "handing you back" is a technical phrase users don't expect to hear. Change to "All done.".

10. "I didn't catch that. You can set timers, alarms, or reminders. Just tell me what you need." — 20 words, reads like a manual. Change to "Didn't catch that — try saying something like 'set a timer for 10 minutes'." — gives a concrete example instead of a category list.

11. "I asked too many questions. Let's start over — just tell me what you need." — exposes internal loop logic to the user. Change to "I'm having trouble with that one. Could you try saying it a different way?"

12. Both error handlers say "Something went wrong. Please try again." — fine for length but offers no path forward on a voice device. Change to "Sorry, something went wrong. Try saying your request again."

[uzair401]

Please address the issues explained in the comments above, and once resolved I'll take another look!

[uzair401]

Approved for community.
We will proceed with adopting this ability for a 30-day evaluation period to monitor performance, identify any bugs, and implement any necessary fixes or optimizations. If the ability proves stable and meets our quality standards during this period, we will move forward with publishing it to the marketplace, ensuring full credit is given to the original author for their work and contribution.
We truly appreciate your effort and encourage you to continue developing and submitting new abilities. Contributions like yours help strengthen the ecosystem, and we look forward to seeing more of your work.

Changes made before merge: Removed hardcoded delete, list, and bare trigger fast-paths. these silently failed on natural/native speech like "get rid of my alarms". Everything now routes through the LLM. Expanded quit/exit/affirmative word lists — critically "stop" was missing. Hardened the LLM prompt to enforce plain spoken English on QUESTION responses. Cleaned up spoken strings and capped list output at 3 items.