SimpleIDGen MCP Server
Plug synthetic US person & patient data straight into your AI agent. Add one server URL to Claude Code, Claude.ai, Cursor, or any MCP client and ask for fake people, diabetic cohorts, or year-by-year patient timelines — returned as CSV, JSONL, or FHIR R4, right in the conversation. Entirely synthetic; no real PII.
The server speaks MCP over Streamable HTTP at https://api.simpleidgen.com/mcp. Authenticate with your clientId as a Bearer token — it's shown in your profile (or returned by GET /v1/auth/me). Create a free account →
# Claude Code
claude mcp add --transport http simpleidgen https://api.simpleidgen.com/mcp \
--header "Authorization: Bearer YOUR_CLIENT_ID"
# Claude.ai / other clients — add a custom connector:
# URL: https://api.simpleidgen.com/mcp
# Header: Authorization: Bearer YOUR_CLIENT_IDThen just ask: "generate 20 fake diabetic patients as FHIR" or "give me a 15-year T2DM timeline for 5 patients as CSV."
| Tool | What it does | Key args |
|---|---|---|
generate_people | Synthetic US adults — demographics, geography, finances, behavior, and calibrated health attributes | count (≤100), seed, format |
generate_t2dm_cohort | Type-2-diabetes patient cohort with staging fields; optional severity dial | count (≤100), stage 1–5, seed, format |
generate_timeline | Longitudinal year-by-year T2DM histories — A1c/eGFR/BMI/BP drift, dated complications, treatment escalation, mortality | count (≤25), horizonYears, seed, format |
describe_schema | The data model — attributes, formats, and where to go for bulk jobs (free, no quota) | — |
format = jsonl (default) · csv · fhir (US-Core-aligned FHIR R4). Same seed ⇒ identical records.
The MCP tools meter against the same free tier as the API — 5,000 rows per UTC day per account. Inline results are capped at 100 people (25 for timelines) to stay chat-sized; for larger or fully-async jobs use the REST API (OpenAPI spec). Your clientId is the bearer credential — keep it private; usage shows in your profile history.