Wellness Cycle Coach

Wellness Cycle Coach

Menstrual cycle coach MCP for AI agents — now with PCOS-aware mode.
Built so AI finally serves the 50% of users agents have ignored — without ever storing cycle data.

[npm version] [npm downloads] [License MIT] [Site]

[GitHub stars] [Built for MCP] [Hermes] [OpenClaw]

🌙 Why this exists: Most AI agents treat the body as a single context-free unit. But energy, recovery, training tolerance, nutrient needs and even cognitive bandwidth shift across the menstrual cycle. wellness-cycle-coach gives any agent phase-aware guidance — and never stores cycle data to do it.

⚡ One-command install — pick your runtime:

• Delx Wellness for Hermes: npx -y delx-wellness-hermes setup
• Delx Wellness for OpenClaw: npx -y delx-wellness-openclaw setup

HTTP (v2 stateless)

Default is stdio. Optional Streamable HTTP — no session id, JSON responses, loopback only:

npx -y [email protected] --http# GET  http://127.0.0.1:3000/health# POST http://127.0.0.1:3000/mcp   (sessionless)

Env: WELLNESS_CYCLE_COACH_HOST, WELLNESS_CYCLE_COACH_PORT, WELLNESS_CYCLE_COACH_TRANSPORT=http.

Overview

Pass in period start dates (from any source — Apple Health Cycle, Garmin women's health, Fitbit female health, or direct user input) and get back the user's current phase plus phase-aware recommendations for nutrition, training, and hydration. Stateless — the MCP itself never persists cycle data. Supports PCOS-aware mode via the cycle_irregular flag (v0.3.3) — accepts cycles 21-90 days, caps confidence at 'low', and returns a luteal_extended placeholder when standard 14-day-luteal math no longer applies.

Try It In 60 Seconds

npx -y [email protected] doctor
# Or use the MCP directly via your client:# {#   "mcpServers": {#     "wellness-cycle-coach": {#       "command": "npx",#       "args": ["-y", "[email protected]"]#     }#   }# }

Then in your agent:

{  "name": "cycle_full_report",  "arguments": {    "history": [      { "start_date": "2026-04-01" },      { "start_date": "2026-04-29" }    ]  }}

Returns current phase + nutrition emphasize/moderate/avoid + training style/intensity + hydration target + next-period estimate.

Tools (17)

The 4-phase model

Why stateless?

Menstrual cycle data is medical-record sensitive. The strongest privacy guarantee is to never store it. Other apps (Flo, Clue) live by hoarding cycle data on their servers; this MCP refuses to participate. The agent passes data in, the coach returns guidance, the data evaporates.

Cross-connector wedge

Apple Health Cycle → period dates       ┐Garmin women's health → cycle context   ├─→ wellness-cycle-coach → phase + guidanceFitbit female health → period dates     ┘                                  │                                                                            │                                                                            ↓                                                            wellness-nourish coach                                                            (phase-aware meal planning)                                                                            │                                                            whoop-mcp / garminmcp / ouramcp                                                            (recovery-aware late-luteal load adjustments)

Privacy

• ✅ Stateless for cycle data — period dates are never persisted; they stay in process memory for the duration of the call and evaporate.
• ✅ Opt-in local preferences — the cycle_profile_* tools can persist non-secret wellness preferences (name, goals, devices, training/nutrition context) to ~/.delx-wellness/profile.json, but only when the user explicitly asks (cycle_profile_update requires explicit_user_intent: true). Secrets (tokens, API keys, biomarkers) are rejected at write time.
• ✅ Offline-capable — pure-function computation. No outbound calls.
• ✅ Tool-arg-only cycle data — the agent passes period history in via the MCP request and it stays in process memory.

Run wellness-cycle-coach doctor to inspect.

What this is NOT

• Not medical advice or diagnosis.
• Not a fertility tracker or contraception aid (consult a clinician).
• Not a replacement for talking to a healthcare provider about painful, abnormal, or absent periods.
• PCOS / irregular cycles supported via cycle_irregular: true (v0.3.3), but this is NOT a substitute for clinical care — see clinician for fertility, contraception, or symptom-management decisions.
• Not specialized for perimenopause or post-pill (yet — see CONTRIBUTING.md).

Roadmap

• v0.2 — adapters for apple-health-mcp / garminmcp / fitbitmcp so agents can pull period history with one MCP call.
• v0.3 — symptom logging surface + symptom-aware guidance adjustments (cramps → magnesium emphasis, mood drop → B-vitamin emphasis).
• v0.4 — non-English locale support starting with pt-BR.

📧 Contact & Support

• 📨 [email protected] — general questions, integration help, partnerships
• 🐛 Bug reports / feature requests — GitHub Issues
• 🐦 Updates — @delx369 on X
• 🌐 Site — wellness.delx.ai

License

MIT — see LICENSE.

_{wellness-cycle-coach is independent research-software. Not affiliated with Clue, Flo, Stardust, or any other cycle-tracking app. Not medical advice.}

Skill or MCP

Same package, two doors. MCP registers tools on stdio/HTTP. The skill can drive the same tools through the CLI when the client has no MCP:

npx -y [email protected] call cycle_connection_status --json '{}'

Copy skill/SKILL.md into your agent skills dir.

0.5.0 — candidate (2026-11-02)

This is a release candidate, not evidence of an npm publication. Existing installations are not updated automatically.

Migration and security boundaries

• HTTP transport accepts loopback only. LAN addresses and wildcard binds are rejected; use stdio or a local loopback client instead.
• Wellness profiles are validated, limited to 256 KiB, and protected with a cooperative lock. Oversized or invalid existing profiles must be corrected before use. Symbolic links, including dangling links and legacy profile paths, are rejected. These checks do not guarantee safety against a hostile local filesystem or non-cooperating processes.
• MCP inputs have a 10,000-element budget and HTTP bodies are limited to 1 MiB; compressed request bodies are rejected. Previously accepted oversized requests may now fail; reduce request sizes rather than disabling validation.
• Dependency fixes are included. Upgrade explicitly after publication and restart the connector; old installations retain their previous behavior.