Tracking your own health symptoms and discovering what triggers them is a challenge many face. The OpenClaw Health Tracker empowers developers and users to automate daily symptom logging, receive reminders, and analyze weekly patterns while preserving data privacy. By integrating OpenClaw with WisGate’s API using Claude Haiku and Sonnet models, you build a local-first AI health symptom tracker that collects natural language input securely on your device and sends only anonymized summaries externally. This guide walks through complete configuration, cron scheduling, system prompts, data storage, and cost transparency for your OpenClaw use cases with WisGate’s affordable platform.
Before you get started, obtain your WisGate API key from https://wisgate.ai/hall/tokens and experiment with prompt designs at https://wisgate.ai/studio/image. This upfront testing helps ensure smooth automation once you deploy your personal health tracking agent.
AI Health Symptom Tracker: Why Patterns Are Invisible Without Consistent Logs
Why is it so hard to find health symptom triggers? Manual logging is often inconsistent, vague, or forgotten altogether. Without consistent and structured symptom records, detecting meaningful correlations is nearly impossible. That’s why automating check-ins with scheduled reminders is critical.
The OpenClaw Health Tracker solves this by prompting you up to three times a day to log your symptoms, food intake, sleep quality, and energy levels in natural language. These logs accumulate locally, creating a rich dataset that a weekly AI analysis summarizes to detect probable triggers and patterns.
Importantly, this is not medical advice. The agent includes mandatory disclaimers stating it’s not a medical advisor. Your raw data stays on your device; only anonymized weekly summaries are sent for analysis to protect your privacy.
Get your WisGate API token at https://wisgate.ai/hall/tokens and validate your prompt templates in AI Studio at https://wisgate.ai/studio/image before fully automating gains.
What the Health Tracker Agent Does
The health tracker agent operates in three main modes:
-
Check-in reminders: Using Claude Haiku 4.5 via WisGate, it sends three daily prompts (morning, afternoon, evening) encouraging natural language symptom reporting.
-
Log ingestion: It parses your natural language entries into JSON objects, encompassing food intake, symptoms, sleep quality, energy levels, and a daily wellness rating.
-
Weekly analysis: Every Sunday morning, Claude Sonnet 4.5 receives anonymized data extracted from locally stored logs for trigger pattern detection.
Privacy is paramount. Raw logs remain in append-only JSON files under ~/.openclaw/health-logs/health-log.json. Only structured, anonymized summaries leave your machine over WisGate’s secure API at https://api.wisgate.ai/v1.
The WisGate custom provider binds these roles: Claude Haiku model for daily check-ins and ingestion, and Claude Sonnet model for weekly trigger analysis.
OpenClaw API Health Tracking Automation: WisGate and Cron Setup
Configuring OpenClaw to use WisGate API involves manually editing the primary JSON config file located at ~/.openclaw/openclaw.json.
Follow these steps:
- Open your terminal and run:
nano ~/.openclaw/openclaw.json
- Insert or replace the models section below inside your configuration, updating the apiKey field with your WisGate API key from https://wisgate.ai/hall/tokens:
"models": {
"mode": "merge",
"providers": {
"moonshot": {
"baseUrl": "https://api.wisgate.ai/v1",
"apiKey": "WISGATE-API-KEY",
"api": "openai-completions",
"models": [
{
"id": "claude-haiku-4-5-20251001",
"name": "Claude Haiku 4.5",
"reasoning": false,
"input": ["text"],
"cost": {"input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0},
"contextWindow": 256000,
"maxTokens": 8192
},
{
"id": "claude-sonnet-4-5",
"name": "Claude Sonnet 4.5",
"reasoning": true,
"input": ["text"],
"cost": {"input": 0, "output": 0, "cacheRead": 0, "cacheWrite": 0},
"contextWindow": 256000,
"maxTokens": 8192
}
]
}
}
}
-
Save the file with Ctrl+O, Enter, then exit nano with Ctrl+X.
-
Restart OpenClaw by pressing Ctrl+C to stop and then run openclaw tui again.
Your cron schedules should be set as follows to capture timely reports without overwhelming yourself:
| Cron Schedule | Purpose | Notes |
|---|---|---|
| 8:00 AM daily | Morning check-in reminder | Captures overnight symptoms, breakfast |
| 1:00 PM daily | Afternoon check-in | Post-lunch symptom and energy logging |
| 7:00 PM daily | Evening check-in | Day recap before sleep |
| 8:00 AM Sunday | Weekly trigger analysis | Processes last 7 days of local logs |
This spread brackets your meals and sleep cycles well.
Check WisGate model docs at https://wisgate.ai/models for latest updates and API options.
LLM Personal Health Agent: Check-in, Logging, and Analysis Prompts
The heart of this AI health symptom tracker is its system prompts, carefully designed for each operational mode.
Check-in Reminders (Claude Haiku): The prompt sends daily reminders urging natural language submission of symptoms, food intake, sleep quality, energy levels, and overall day rating — respecting privacy and including mandatory disclaimers repeated thrice: “This agent is not a medical advisor.”
Natural Language Log Ingestion: After user entry, Claude Haiku parses entries into a strict JSON schema:
- "date": ISO 8601 date of entry
- "food": description of meals/snacks
- "symptoms": free text symptom descriptions
- "sleepQuality": subjective measure
- "energyLevel": subjective energy rating
- "dailyRating": overall day rating 1-10
This structure keeps logs consistent for weekly analysis.
Weekly Trigger Analysis (Claude Sonnet): On Sundays, Claude Sonnet receives anonymized weekly data extracted from local logs for pattern detection. The prompt includes explicit disclaimers emphasizing no medical advice is given and details the analysis scope.
The prompt roles enforce clear boundaries and parsing instructions, improving the quality of insights and protecting both user and developer.
Local Log Schema and Weekly Report Delivery
All raw user entries are stored locally and appended as JSON objects in a file at ~/.openclaw/health-logs/health-log.json. This append-only format ensures auditability and immutability.
Weekly after the Sunday analysis, an anonymized Markdown report is generated and saved as ~/.openclaw/health-logs/reports/week-[DATE].md, ready for review or optional email delivery if configured.
To prevent model context limit overflow (256,000 tokens max), logs older than 90 days are rotated out — either archived or deleted. This keeps the analysis relevant to the recent period and manageable for Claude models.
OpenClaw Use Cases: Cost Per Week of Health Tracking
Understanding token consumption and costs is critical for any AI integration project.
Here’s a breakdown of expected weekly token usage across the scheduled calls:
| Model ID | Calls Per Week | Tokens Per Call | Total Tokens Weekly |
|---|---|---|---|
| claude-haiku-4-5-20251001 | 21 (3/day × 7) | ~400 | 8,400 |
| claude-haiku-4-5-20251001 | 1 (log ingestion) | ~6,300 | 6,300 |
| claude-sonnet-4-5 | 1 (weekly analysis) | ~4,800 | 4,800 |
| Total Weekly Tokens | ~19,500 tokens |
Annual consumption is approximately 1.01 million tokens, combining all model calls.
For cost context, WisGate’s image generation pricing is $0.058 per stable quality image, while the official rate is $0.068—showing WisGate’s competitive pricing. Token prices vary, so check https://wisgate.ai/models for latest pricing.
Because raw logs stay local and only summaries reach WisGate, this budget covers AI compute costs efficiently.
OpenClaw Use Cases: Consistent Logging, Local Data, Weekly Insights
Deploying this OpenClaw configuration starts with editing ~/.openclaw/openclaw.json to register WisGate models, setting your API key, and establishing cron jobs for reminder prompts and weekly analysis.
Typically, within one to two weeks of consistent logging, you’ll begin to see meaningful trigger patterns emerge in weekly reports.
Your raw symptom and food intake logs never leave your machine, ensuring full data residency with only structured anonymized summaries transmitted.
This agent is designed as a developer tool for personal or experimental health tracking—not for medical diagnostics or clinical advice. Multiple disclaimers appear in prompts and documentation emphasizing that.
Get your WisGate API key at https://wisgate.ai/hall/tokens and test your prompt behavior at https://wisgate.ai/studio/image to refine recollected data.
Remember: three check-in reminders daily spaced around your meals and sleep help keep data consistent without being intrusive. Cron jobs enable this schedule without extra effort.
With WisGate, you access top-tier models for check-in and analysis under one affordable API, building your personal AI health symptom tracker quickly and transparently.
Your check-in cron schedule is set, system prompts are ready, and WisGate API is integrated. Now, obtain your WisGate API key at the tokens page, validate your prompts in AI Studio, and start consistent logging for insightful weekly health trigger analysis—all with no sensitive data leaving your machine except anonymized reports.
Links:
Tags: OpenClaw, Health Tracking, WisGate API
Social: Configure OpenClaw with WisGate’s API for privacy-first AI health symptom tracking and weekly trigger detection.