Shipped: Fixed chat window overlapping bottom buttons in Voice Orchestrator iOS app.

Added bottom padding to the ChatView panel so the message input field no longer clips into the MIC/VOICE buttons.

Next: Continue polishing the UI.

Health — 2026-03-04

1,436 steps

Notes

I imagine having an addon that auto assigns users based on their class, spec, and role to predefined templates for boss fights.

For example, high king in wow tbc.

Assign:

• tanks to bosses

• Healers to tanks

• Auto mark bosses

Announce:

• assignments

• Kill order

• Priority

• Mage tank

This would also generate announcements that explain:

• boss mechanics (I.e. melee watch out for maulgar whirlwind and run out, don’t dispel mage boss, etc.)

The addon should do its best to automate these aspects, but it should also give me the ability to customize it. For example healer A was assigned o heal tank A, but I decided to make them to tank B. The addon interface supports and persists this and the announcements adjust to reflect these changes within the addon.

Now this should be added to each raid and each boss. Tons of trash mobs in raids are unique and dangerous, so ideally this is announced and explained too.

The goal is to be able to run a raid purely via chat and not need discord or alternative applications. I want these assignments and announcements to be explicit and effective.

This addon should also help users create the raid, bringing in the correct number or tanks healers melee and ranged dps.

Bonus points if we can package it like RestedXP, selling raid packs.

Organic marketing by reading successful pugs using it to announce and appending the addon name on announcement.

Voice agent is the orchestration layer with:

Tools:

• spawn: Able to spawn subagents that are fired off to work on particular coding tasks.

  • Params:

    • prompt

    • identifier

      • Identifier should be a natural key that is easily identifiable given the context of our check prompt

    • number (optional)

      • Should support spawning many sub agents for a single task

  • Example: “build me a website about …”

• check: manual peak into the state of the subagent.

  • Params:

    • agent identifier

  • Example: “how is our web builder agent doing?”

Skills:

• GitHub CLI

Open questions:

• How does the orchestration layer know when a subagent is finished with its task?

• What if multiple subagents finish at the same time?

  • Do we prioritize based on a certain criteria?

  • Queue them one by one?

• How do we manage our voice agent’s context relative to the spawned tasks?

  • Ideally we do not need to start a new voice session for each task we are working on.

Thoughts:

• I like the idea of the voice agent being my interface into the computer. It should be generic and open-ended.

How can I pipe server and console logs to my agent?

I deploy my app and the deployment breaks - I want this context to get back to my agent so it continues working on it until this passes.

I can use OpenClaw to trigger on events. I.e. on deployment fail, spawn a session with the context of the message.

But… I think it may be better to fix it in the environment that did the deployment. Maybe this is incorrect and subagents are perfectly fit for this task.

Built AlwaysOnTop — a tiny macOS menu bar utility that pins any window above everything else.

Hit Right⌘ + Right⌥ + / on any focused window and it floats above all other windows, even fullscreen apps. Uses ScreenCaptureKit to mirror the window into a floating overlay at screenSaver level.

Had to reach for some private CoreGraphics APIs (CGSSetWindowLevel, _AXUIElementGetWindow) since modern macOS doesn't let you manipulate window levels for other apps through public APIs. The overlay approach works around this by capturing and redisplaying the window content.

Useful when you want to keep a video, reference doc, or terminal visible while working in another app.

Built a pi skill to instantly pull my latest screenshot into context. Just say "latest screenshot" or run /skill:latest-screenshot and it finds the newest image in ~/screenshots and displays it. Also supports grabbing the last N screenshots (e.g. /skill:latest-screenshot 3). Simple shell script under the hood — sorts by modification time across png/jpg/gif/webp/tiff.

Health.md v1.5.0 submitted for both iOS and macOS.

New in this release: quickly switch all frontmatter keys between snake_case and camelCase right from the Frontmatter Fields settings. Added a key style picker and toolbar shortcut for fast switching, plus minor improvements and bug fixes.

Both platforms archived, uploaded, and submitted for App Store review in one session.

Health.md for macOS is now live on the App Store! 🎉

Shipped: Replaced all Gumroad links across the iOS app, iPad app, and marketing website to point to the new Mac App Store listing.

Decided: Single distribution channel through the App Store for both iOS and macOS — no more Gumroad for the Mac app.

Next: Push the updated website live and submit the iOS/iPad app update with the new links.

https://apps.apple.com/us/app/health-md/id6757763969

Shipped the macOS version of imghost to App Store Connect.

Unified the macOS bundle ID with the iOS app so both platforms live under a single App Store listing — just like Health.md. Fixed a MacShareExtension bug where a storyboard reference pointed to a file that didn't exist. Binary uploaded, metadata synced, build selected.

Still need screenshots before submitting for review.