Personal AI Hub: Apple Health + Threads + Gmail + Calendar
Build a single Telegram chat with a Hermes agent that pulls your Apple Health sleep, Threads analytics, Gmail inbox and Google Calendar on demand and cross-references them in a daily morning brief.
Requirements
- The deployable artifact is
krumjahn/threads-growth-skillon GitHub — an open-source Hermes skill that pulls Threads analytics into the agent. Other skills in the workflow are documented in the blog but not all are open-sourced. - Apple Health data must be exported to a local file, using the Health Auto Export app or a custom iOS Shortcut; the analyzer reads the JSON and persists it as a daily skill.
- Threads access uses
krumjahn/threads-growth-skillfrom GitHub with authentication via exported browser cookies (Cookie-Editor extension) — no API key needed. - A Google Cloud OAuth app with the Gmail API and Calendar API enabled, with the credentials JSON placed in
~/.hermes/integrations/google/. - A shared knowledge base: the blog uses an Obsidian vault at
~/ObsidianVault/, which Hermes reads and writes directly. krumjahn/threads-growth-skill(open-source, deployable).- Apple Health export pipeline (Health Auto Export app on iOS).
- Google Cloud OAuth (Gmail + Calendar).
- Telegram gateway with Topics.
- Obsidian vault (optional, for shared knowledge base).
- An LLM provider — the blog uses OpenAI Codex ($20/mo), OpenRouter (free tier), or local Ollama.
- Cost: the blog uses OpenAI Codex ($20/mo) or OpenRouter's free tier; both are sufficient for the morning brief workload.
- Deployable skill:
krumjahn/threads-growth-skill— open-source Hermes skill for Threads analytics.
How to use
This setup suits a solo founder or technical operator who wants a single chat surface to query personal data across multiple services, with a long-running agent that accumulates knowledge over time.
Browser-cookie auth for Threads is the path that works today, since the Threads API is restricted. The cookie approach is in the same gray area as other browser-cookie integrations — fine for personal use, fragile if Meta changes the export format.
Cross-domain queries only work if the agent has been writing to a shared knowledge layer (Obsidian, or a structured memory skill); the setup is only as durable as the write-through to that layer.
- Install Hermes Agent on your main machine (the blog uses macOS; Linux/Windows work the same with platform-appropriate substitutions).
- Configure Telegram as the primary gateway and enable Topics.
hermes gateway setup
# pick Telegram, paste bot token, set chat ID with Topics enabled
- Wire each service. Apple Health: export Health data to a local file (Health Auto Export app, or a custom iOS Shortcut) and point Hermes at the export path; the analyzer reads the JSON and persists it as a daily skill. Threads: install
krumjahn/threads-growth-skillfrom GitHub; auth is via exported browser cookies (Cookie-Editor extension) — no API key needed.
```bash
git clone https://github.com/krumjahn/threads-growth-skill.git
cp -r threads-growth-skill ~/.hermes/skills/threads-growth-skill
hermes skills reload
```
Gmail + Google Calendar: create an OAuth app in Google Cloud Console with Gmail API and Calendar API enabled, then drop the credentials JSON into ~/.hermes/integrations/google/.
- Add a daily morning cron that pulls sleep + calendar + inbox + Threads into one brief.
/cron add "0 7 * * *" "Every morning at 7am: read my sleep data from
~/.hermes/data/health/last-7-days.json, my calendar for the next 24h,
my unread Gmail (inbox-only, max 20), and my Threads analytics for the
past 7 days. Compose a 5-bullet morning brief: sleep, schedule, top 3
emails, top Threads post, and a one-line suggestion for the day.
Post to my Telegram health topic."
- Set up the shared knowledge base: the blog uses an Obsidian vault at
~/ObsidianVault/, and Hermes reads/writes notes there directly.
Prompts.
# Health
"Show me my average sleep this week vs. last week, and call out any
night under 6 hours."
# Social
"Pull my top 3 Threads posts from the last 7 days by impressions,
with their first 50 characters of copy."
# Cross-domain
"On days I shipped a release in the last 30 days, what was my average
sleep the night before?"
# Inbox
"Show me unread inbox messages from senders I've replied to before
and that arrived after 6pm yesterday."
The cross-domain query is the killer feature — it requires Hermes to know about both Health data and release history, which the shared Obsidian vault enables.
Links
Source: aliaihub/awesome-hermes-usecases and its contributors. License: CC-BY-4.0. Source license notice.
Translation and editorial adaptation by ClawCases.
>Resources
>Related Use Cases
AI-Powered Earnings Tracker
Stay on top of earnings season across dozens of tech companies without manually checking multiple sources and remembering report dates for every AI/tech earnings report.
AI Video Editing via Chat
Describe your edit to an agent to trim videos, add music, subtitles, or crops — no timeline or GUI. Install two skills, then editing is done by conversation.
Semantic Memory Search with memsearch
Add vector-powered semantic search on top of OpenClaw's plain-markdown memory files with memsearch, so you can instantly find any past memory by meaning instead of grepping keywords or loading whole files into context.