Skip to content

v1.47.6

Latest

Choose a tag to compare

@MBombeck MBombeck released this 15 Nov 10:58
· 17 commits to main since this release
93f6891

A small patch made of your requests. Medications get a note for the things you otherwise keep on a sticky note by the pill box, home automations can log custom metrics, and if you ever tried to log an illness before lunch, HealthLog will now believe you're sick in the morning too.

A note on a medication

Directions such as "30 to 60 minutes before food" have a place of their own now. Type them into the last step of the medication wizard, and they show under the medication's name on its page, above the Taken and Skip buttons on its card, and under the medication when you log a dose. The note is up to 500 characters, encrypted like your other notes, and goes along with backups, restores and the data export. (#1371)

Custom metrics from scripts and Home Assistant

A measurement token can now post values to a custom metric you already created, through POST /api/custom-metrics/{id}/entries. It still can't list, read, change or delete anything, and a metric of another account answers 404. The metric's ID is the part after /custom-metrics/ in its page address, and the token card in Settings says so. Tokens you already have can do this too. (#1377)

Fixed

  • Creating an illness episode in the morning works again. The form sent the start date as noon, which before noon is in the future, so the save was refused. Start and end dates also stay on the day you picked when your time zone is far from UTC, and editing an episode keeps the times you saved. (#1376)
  • On a phone, "Open a record" in the account menu now unfolds the records inside the menu itself instead of opening a second menu off the edge of the screen. (#1372)
  • When the browser's push service refuses a registration, Browser Push now says so and links the help page instead of a bare "Activation failed". In Brave it names the setting to turn on. (Discussion #1378)
  • The nightly "insight-pregenerate still failing" alert no longer fires over an AI provider that can't answer. An account with a provider still saved whose key the provider refuses, or whose providers HealthLog paused after repeated errors, used to count as a failed run every night. Removing only the key from a Local or OpenAI-compatible provider keeps its address, so that provider was still being called. The Background jobs card on the admin status page now lists the same runs the alert counts, and both say which provider failed and why. (#1384)
  • Switching into or out of a shared record while HealthLog was open in a second tab could leave both tabs on the loading spinner for half a minute. An account check that started just before the switch could answer for the record being left and undo it. Such an answer is now thrown away and asked again.
  • Charts no longer jump while they load. Most chart placeholders were 20 px taller than the chart they turned into (40 px on a phone) and the mood chart's was about 30 px shorter, so the page shifted under you when the charts arrived.

Security

The hono floor moves to 4.13.16 (CVE-2026-126355 to CVE-2026-126358). HealthLog only gets it through the MCP SDK and doesn't load the affected parts, so this clears the audit rather than closing a reachable hole.

Updating

Pull and restart. One small migration runs on the first start and adds the medication note column.

Thank you

@lbradley-mm asked for the medication note and found the record menu running off a phone screen (#1371, #1372). @kruemelro wanted their custom metrics fed from home automation (#1377). @oilee80 hit the silent push failure in Brave and stayed with it until we knew why (#1378). @ferges123 sent the illness fix as a pull request and trimmed it down without a word of complaint (#1376). @mchal reported the nightly job that cried wolf (#1384), and @mills1975 kept testing the iPhone app's sync for a week.

And thanks to everyone who uses the iPhone app, sends an email or buys me a coffee. It keeps this going.