Purpose
3AM Decoder MCP server is provided by The Longevity Vault and is used to identify the possible biological causes of waking at 3 AM based on the sleep symptoms described by the user. It is officially emphasized that most sleep advice targets only symptoms, while this server classifies the biological mechanisms behind sleep disruption into one of five causes.
Main Capabilities
The server exposes the following tools:
- classify_sleep_cause: Classifies the likely main cause of sleep disruption based on the symptom description. Input symptoms (string, required, up to 2000 characters), which should include what happens (such as waking at 3 AM, difficulty falling asleep), duration, age, sex, and relevant context (gut issues, hot flashes, eating patterns, etc.). Output is the main cause and explanation, confidence, and a link to the full interactive Decoder for comprehensive multi-cause analysis.
- Detailed cause lookup: When the user already knows or has classified a cause, provides a deeper explanation of the biological mechanism. Input cause_id (enum, required), with possible values autonomic, metabolic, inflammatory, hormonal, circadian. Output is the full mechanism explanation, observable symptoms, and what people usually notice.
- List all 5 causes of sleep disruption and summaries: Used when the user wants to first understand the overall framework or explain the biology of sleep disruption.
- get_decoder_url: Get a tracking URL pointing to the full interactive 3AM Decoder. Optional parameter utm_campaign (string) for analytics tracking. Output is a tracking URL with UTM parameters. The Decoder is free, takes about 3 minutes, and can map primary, secondary, and tertiary causes.
The classifier uses weighted keyword matching and applies age/sex adjustments derived from the same domain knowledge.
How to Use
It can be connected through the Smithery gateway, or directly to the Cloudflare Workers endpoint, with Streamable HTTP (stateless) transport. The server can be added to claude_desktop_config.json, or to the MCP settings in Cursor / VS Code. An official example input for classify_sleep_cause is provided: "52 year old woman, waking at 3am for the past 2 years. Hot flashes, night sweats, needing to pee 2-3 times. Energy is lower than it used to be."
Limits: 100 requests per minute per IP; symptom input maximum 2000 characters; no authentication required.
Applicable Scenarios
Useful when a user describes a sleep problem and wants to understand its biological cause (rather than receive generic sleep hygiene tips); or when the user already knows the cause and wants a deeper understanding of the mechanism; it can also be used to give the user an overview of the five-cause framework.
Important Notes
This classifier provides a simplified identification of the main cause. Most people over 40 have multiple causes, and the combination is often more important than a single cause. The full 3AM Decoder provides comprehensive multi-cause analysis through a 15-question weighted scoring engine. It is officially stated that the classifier achieves 92% accuracy on test scenarios derived from the full Decoder scoring engine (34/37 passed; 3 failures were true edge cases). The sleep disruption cause framework was developed by Kat Fu, M.S., M.S.
