Analysis updated 2026-05-18
Generate a written report of what happened during a specific sleep episode from Garmin and Minowa data.
Build up a growing log of what pre sleep habits led to well scored versus poorly scored nights.
Let a Claude skill correct and remember your personal health inference rules over time.
| minowahealth/nightwatch | abhas9/escape-run | abhishek-kumar09/mern-live-js-june-2021 | |
|---|---|---|---|
| Stars | 1 | 1 | 1 |
| Language | HTML | HTML | HTML |
| Last pushed | — | — | 2021-09-13 |
| Maintenance | — | — | Dormant |
| Setup difficulty | moderate | easy | moderate |
| Complexity | 2/5 | 2/5 | 2/5 |
| Audience | general | general | developer |
Figures from each repo's GitHub metadata at analysis time.
Requires a connected Minowa MCP server and a Garmin device already syncing sleep and HR data.
NightWatch is a pair of Claude Skills, meaning add on capabilities for Anthropic's Claude AI assistant, built for people who track their sleep and health with a Garmin wearable device and log additional information like blood pressure and medications through a personal health tracking system called Minowa. Rather than dumping raw numbers on the user, both skills turn that tracked data into written analysis. The first skill, health episode report, answers the question of what happened during a specific stretch of time, such as one night or a single nap, by pulling heart rate, stress, sleep stages, blood pressure, and any manually logged notes for that window and writing them up as a single page report. The second skill, scored sleep signature, looks specifically at nights that Garmin was able to fully score, and reconstructs what the person did in the hours leading up to that night, so patterns in what leads to a good or a mediocre night can build up and be compared over time. Both skills rely on a Minowa MCP server already being connected to Claude to fetch the underlying health data, and can be installed either as a packaged skill file or directly from the source folders in this repository, in which case the second skill also needs the first one installed alongside it. A notable feature is that both skills keep simple text log files that accumulate corrections over time. When a person tells the assistant it misread a piece of data, the correction gets written down and reused going forward, so the same mistake should not repeat. Because those log files can contain real personal health details like medication names and blood pressure readings, the README explicitly warns users to think about whether to keep their copy of the repository private, and provides example template files without personal data for anyone who wants to share the code publicly. The project is released under the MIT license and states clearly that it is not medical advice.
Two Claude Skills that turn Garmin and Minowa health tracking data into written sleep and health episode reports that improve from corrections.
Mainly HTML. The stack also includes Claude Skills, Minowa MCP, HTML.
Setup difficulty is rated moderate, with roughly 30min to a first successful run.
Mainly general.
This repo across BitVibe Labs
Verify against the repo before relying on details.