Surveys all connected health data sources (Fasten, HealthEx, MEDENT, etc.) and returns connection status plus record counts by type. Answers 'what records do you have?' without revealing clinical values.
How this skill is triggered — by the user, by Claude, or both
Slash command
/healthclaw-guardrails:check-sourcesThe summary Claude sees in its skill listing — used to decide when to auto-load this skill
Give the patient a one-shot survey of every health data source HealthClaw can
Give the patient a one-shot survey of every health data source HealthClaw can pull from — which ones are connected, when each last had activity, and how many records of each type are on file. This answers "what do you actually have?" in a single call instead of probing each integration one at a time.
The seven sources surveyed:
These apply always. Not overridable by persona or patient request.
sources_check returns connection status and record counts by type. That is
the entire surface you may present from this skill. You may say "57 Conditions,
120 Observations." You may NOT name a specific condition, lab value, medication,
or any other clinical content. To discuss what a record actually says, use
fhir_search / fhir_read (separately audited and redacted) — never infer
clinical detail from this summary.
Report only what sources_check returns. If a source shows connected: false,
say it is not connected and offer to help connect it — do not imply data exists
that the response does not show.
Public/demo tenants (e.g. desktop-demo) work without a token. For protected
tenants (e.g. ev-personal), the command-center endpoint requires step-up auth,
so get a token first:
fhir_get_token(tenant_id: "<tenant>")
→ returns { token: "..." }
sources_check({ _stepUpToken: "<token from step 1>" })
sources_check takes no required arguments — the tenant comes from the request
context. Pass _stepUpToken so it forwards to the protected endpoint. If you
omit it on a protected tenant the call returns requires_step_up: true; get a
token (Step 1) and retry.
The response includes:
connected_count / source_count — e.g. 3 of 7sources[] — each { id, name, connected, detail, last_activity }total_recordsrecords_by_type[] — each { type, count }_mcp_summary — e.g. "3 of 7 sources connected; 177 total records."Present it tidily:
Connected: 3/7 — Fasten ✓, MEDENT ✓, Wearables ✓
Not connected: HealthEx, Health Bank One, Flexpa, Epic/Health Skillz
You have 177 records on file:
• 57 Conditions
• 120 Observations
fhir_search)Patient: What's connected? Do you actually have my records?
Sally: Let me check all your data sources at once.
[fhir_get_token(tenant_id: "ev-personal") → { token: "..." }]
[sources_check({ _stepUpToken: "..." }) → {
connected_count: 3, source_count: 7, total_records: 177,
sources: [
{ id: "fasten", name: "Fasten", connected: true },
{ id: "medent", name: "MEDENT", connected: true },
{ id: "wearables", name: "Wearables", connected: true },
{ id: "healthex", name: "HealthEx", connected: false },
...
],
records_by_type: [
{ type: "Condition", count: 57 },
{ type: "Observation", count: 120 }
]
}]
Connected: 3 of 7 sources — Fasten, MEDENT, and your wearables.
Not connected yet: HealthEx, Health Bank One, Flexpa, and
Epic/Health Skillz.
You have 177 records on file: 57 Conditions and 120 Observations.
Want me to pull anything from one of these, connect another
source, or open your dashboard?
fhir_search / fhir_read.connected: true with 0 records means the link is live but
no data has arrived yet — say so rather than implying records exist._stepUpToken, the call returns
requires_step_up: true; get a token and retry.npx claudepluginhub aks129/healthclawguardrails --plugin healthclaw-guardrailsConnects to US health systems via HealthEx to pull clinical history, analyze with Claude, and optionally export to a curated FHIR store. Triggers on health record requests.
Imports and standardizes health data from Apple Health, Fitbit, Oura Ring, CSV/JSON; integrates WellAlly.tech knowledge base for data management and personalized article recommendations.
Shows freshness of health data sources (Apple Health, Oura, Fitbit, labs), last coach prescription and completion, missed-day count, lab status flags, and auto-trigger hook installation status for debugging the health-mcp auto-trigger chain.