Health Insights — Unified Entry Point

Plugin for personal health data ingestion, baseline establishment, trend analysis, and insight generation. Feeds into MongoDB (primary store) and optionally an Obsidian vault (downstream archive).

All other health-insights skills are internal (triggered by Adam scheduling or agents, not directly by users).

Arguments

Parse from user input (natural language routing):

User says	Route to	Description
/health or /health status	health-ingest-agent (lightweight)	Check last sync state
/health ingest	health-ingest-agent	Adam trigger: parse XML delta to MongoDB
/health baseline	health-baseline-agent	Compute/update personal baselines
/health analyze <topic>	health-analyze-agent	Generate trend insights (includes calendar context enrichment when mactools available)
/health analyze --weekly	health-analyze-agent	Generate weekly correlation analysis
/health digest	health-analyze-agent	Weekly digest: Get笔记 save + IEF export + calendar context
/health predict	health-predict-agent	Run early warning evaluation (triggered daily by Adam cron; alerts delivered to WeChat)
/health report <file>	health-report-agent	Parse lab report
/health trends <type>	health-analyze-agent	Query specific metric trends
/health annual <year>	health-analyze-agent	Generate annual health report
/health setup-delivery	—	Print Adam delivery rule setup instructions
/health --help	—	Show this routing table

Routes

Route: Status (default)

Query MongoDB for current state:

1. db.metrics.count_documents({}) for total record count
2. db.metrics.find_one(sort=[("timestamp", -1)]) for last ingest timestamp
3. db.baselines.count_documents({}) for baseline count
4. db.alerts.count_documents({"status": "active"}) for active alert count

Return a compact status:

Health Insights Status
  Records: 3,247,891
  Date range: 2024-03-15 → 2026-04-09
  Baselines: 24 computed
  Active alerts: 2

Route: Ingest

Trigger: Adam watch folder event or manual ingest arg.

Dispatch health-ingest-agent with:

input:
  source: "{path to XML file or iCloud delta directory}"
  resume_from_byte: 0
  processing_state: {}

Route: Baseline

Trigger: /health baseline or Adam daily schedule.

Dispatch health-baseline-agent:

input:
  action: "compute"   # or "update" for incremental
  metric_type: null    # null = all metrics
  mongo_uri: null      # defaults to $MDB_MCP_CONNECTION_STRING
  database: "health"
  notion_trends_db_id: "<notion_database_ids.trends from config>"
  notion_alerts_db_id: "<notion_database_ids.alerts from config>"

Route: Analyze

Trigger: /health analyze <topic>.

Dispatch health-analyze-agent:

input:
  action: "daily"   # or "weekly", "trend", "annual"
  date: "{YYYY-MM-DD}"
  topic: "{metric type or null}"
  year: null        # for annual
  mongo_uri: null
  database: "health"
  notion_trends_db_id: "<notion_database_ids.trends from config>"
  notion_workspace_page_id: "<notion_workspace_page_id from config>"
  # Note: calendar context enrichment is applied automatically when mactools is available

Route: Predict

Trigger: /health predict or Adam evening schedule.

Dispatch health-predict-agent:

input:
  action: "evaluate"   # or "acknowledge"
  alert_id: null
  mongo_uri: null
  database: "health"
  notion_alerts_db_id: "<notion_database_ids.alerts from config>"

Route: Report

Trigger: /health report <file path>.

Dispatch health-report-agent:

input:
  file_path: "/path/to/lab-report.pdf"
  file_type: null   # null = auto-detected from extension; "pdf" | "image" | "text"
  notion_reports_db_id: "<notion_database_ids.reports from config>"
  notion_lab_results_db_id: "<notion_database_ids.lab_results from config>"

Route: Annual

Trigger: /health annual <year>.

Dispatch health-analyze-agent:

input:
  action: "annual"
  year: "{YYYY}"
  notion_workspace_page_id: "<notion_workspace_page_id from config>"

Route: Digest

Trigger: /health digest.

Dispatches the weekly-digest action which saves a health summary to Get笔记 and exports an IEF insight file for pkos.

Dispatch health-analyze-agent:

input:
  action: "weekly-digest"
  database: "health"
  notion_trends_db_id: "<notion_database_ids.trends from config>"

Route: Setup-Delivery

Trigger: /health setup-delivery.

Prints these setup instructions:

=== Adam Delivery Rule Setup ===
Health Insights uses Adam task templates (config/adam-task-templates.yaml) and
delivery rules to push summaries and alerts to WeChat.

1. Import templates: copy config/adam-task-templates.yaml to Adam's config dir
2. Create delivery rules via Adam Web UI or API:
   POST /delivery-rules

   Rule A — Daily summary to WeChat:
     name: health-daily-summary
     event: task_complete
     matchCriteria: { templateId: "health-daily-analyze" }
     targetChannel: <your_wechat_channel_id>
     skipOriginChannel: true

   Rule B — Alert push to WeChat:
     name: health-alert-to-wechat
     event: task_complete
     matchCriteria: { templateId: "health-predict" }
     targetChannel: <your_wechat_channel_id>
     skipOriginChannel: false

See: config/adam-task-templates.yaml for full API examples.

Data Storage

Primary store: MongoDB health database (metrics, baselines, alerts, lab_reports, checkpoint collections).

Primary insights archive: Notion Health workspace (Trends DB, Alerts DB, Reports DB, Lab Results DB).

Optional downstream archive: Obsidian vault at ~/Obsidian/HealthVault/ (not required for operation; Notion is the primary record for analyzed data).

Cross-Plugin Integration

Target	Mechanism	Frequency	Data
Get笔记	getnote.sh save_note (via getnote_digest.sh)	Weekly (Monday 9am)	Aggregate health summary (~200 chars)
pkos	IEF markdown file export to ~/.adam/state/health-insights/ief-exports/	Weekly	Health insight narrative; pkos intel-sync reads from that path
WeChat	Adam delivery rule (task_complete event)	Daily + on alert	Summary + alert text
Calendar	osascript query to Apple Calendar	On analyze	Event context for correlation (travel, exercise, overtime, social)

Agent I/O Convention

All agents receive YAML input and return YAML output via their tool definitions. The entry skill handles natural language → YAML routing.