Is patient-panel-overview safe to install?

SaferSkills independently audited patient-panel-overview (Agent Skill) and scored it 100/100 (green). The audit ran 55 deterministic rules across Security, Supply Chain, Maintenance, Transparency, and Community; it found 0 high-severity and 0 lower-severity findings. The full rule-by-rule trace and per-finding evidence are below. Free, methodology-open.

CategoryWeightCategory scoreContribution

Securityprompt, exec, net, exfil, eval

35%

100

35.0 pts

Supply chainhash, typosquat, maintainer, lockfile

20%

100

20.0 pts

Maintenancestaleness, pinning, CI

15%

100

15.0 pts

TransparencySKILL.md, perms, README

15%

100

15.0 pts

Communityinstalls, verify, response

15%

100

15.0 pts

Findings & checks · 0 flagged

Securityscore 100 · 0 findings

✓—All security checks passedNo findings in this category for the latest scan.pass

Supply chainscore 100 · 0 findings

✓—All supply chain checks passedNo findings in this category for the latest scan.pass

Maintenancescore 100 · 0 findings

✓—All maintenance checks passedNo findings in this category for the latest scan.pass

Transparencyscore 100 · 0 findings

✓—All transparency checks passedNo findings in this category for the latest scan.pass

Communityscore 100 · 0 findings

✓—All community checks passedNo findings in this category for the latest scan.pass

The primary manifest — the file an agent reads to learn what this artifact does.

skills/core/population-health/patient-panel-overview/SKILL.md · 1 file

skills/core/population-health/patient-panel-overview/SKILL.md12.1 KB · Markdown

name: patient-panel-overview

description: |

Queries patient cohorts by condition and generates panel-level analytics including aggregate metrics, risk stratification, and outreach priorities. Use when user asks to "show my panel", "list all diabetics", "how many patients have CHF", "panel overview", "cohort summary", "population metrics", "who needs outreach", or mentions "panel management", "risk stratification", "quality metrics". Do NOT use for individual patient chart review, single-patient lab interpretation, or quality measure calculation (use quality-measure-dashboard instead).

metadata:

author: LangCare

version: 1.0.0

mcp-server: langcare-mcp-fhir

category: population-health

Patient Panel Overview

Overview

Query patient populations by chronic condition, aggregate clinical metrics across the cohort, stratify patients by risk level, and identify those needing priority outreach. Supports common chronic disease panels: diabetes, hypertension, heart failure, COPD, asthma, CKD, and depression. Produces summary statistics with actionable outreach lists.

FHIR Resources Used

Resource	Purpose	Key Fields
Condition	Identify patients with target condition	code, clinicalStatus, subject, onsetDateTime
Patient	Demographics, age calculation	birthDate, gender, name, identifier
Observation	Lab values for aggregate metrics	code, valueQuantity, effectiveDateTime, status
Encounter	Recent visit tracking	class, status, period, subject
MedicationRequest	Treatment pattern analysis	medicationCodeableConcept, status, subject
Goal	Patient goal tracking	lifecycleStatus, target, achievementStatus

Instructions

Step 1: Identify Target Cohort

Query patients with the target condition using SNOMED CT or ICD-10 codes. See references/panel-management.md for condition-specific codes.

For diabetes panel:

Tool: fhir_search
resourceType: "Condition"
queryParams: "code=http://snomed.info/sct|44054006&clinical-status=active&_count=200"

SNOMED 44054006 = Type 2 diabetes mellitus. Use _count=200 and paginate via _offset if needed.

For hypertension panel:

Tool: fhir_search
resourceType: "Condition"
queryParams: "code=http://snomed.info/sct|38341003&clinical-status=active&_count=200"

SNOMED 38341003 = Hypertensive disorder.

For CHF panel:

Tool: fhir_search
resourceType: "Condition"
queryParams: "code=http://snomed.info/sct|42343007&clinical-status=active&_count=200"

SNOMED 42343007 = Congestive heart failure.

If SNOMED search returns zero results, retry with ICD-10:

Tool: fhir_search
resourceType: "Condition"
queryParams: "code=http://hl7.org/fhir/sid/icd-10-cm|E11&clinical-status=active&_count=200"

Extract unique patient references from Condition.subject. Build a list of patient IDs.

Step 2: Retrieve Patient Demographics for Cohort

For each patient ID in the cohort (batch where possible):

Tool: fhir_search
resourceType: "Patient"
queryParams: "_id=[id1],[id2],[id3],...&_count=50"

Extract: name, birthDate (calculate age), gender, address (for geographic analysis if requested).

Step 3: Pull Aggregate Lab Values

Query condition-specific lab observations across the cohort. See references/panel-management.md for LOINC codes per condition.

Diabetes -- HbA1c:

Tool: fhir_search
resourceType: "Observation"
queryParams: "code=http://loinc.org|4548-4&date=ge[12-months-ago]&_sort=-date&_count=200"

LOINC 4548-4 = Hemoglobin A1c/Hemoglobin.total in Blood.

Hypertension -- Blood Pressure:

Tool: fhir_search
resourceType: "Observation"
queryParams: "code=http://loinc.org|85354-9&date=ge[12-months-ago]&_sort=-date&_count=200"

LOINC 85354-9 = Blood pressure panel.

CKD -- eGFR:

Tool: fhir_search
resourceType: "Observation"
queryParams: "code=http://loinc.org|33914-3&date=ge[12-months-ago]&_sort=-date&_count=200"

LOINC 33914-3 = Glomerular filtration rate/1.73 sq M.predicted.

Filter results to only patients in the cohort. Take most recent value per patient.

Step 4: Check Recent Visit Status

Tool: fhir_search
resourceType: "Encounter"
queryParams: "date=ge[6-months-ago]&status=finished&class=AMB&_count=200"

Match encounters to cohort patients. Identify patients with no visit in the last 6 months.

Step 5: Compute Panel Metrics

For each condition panel, calculate:

Panel size: Total patients with active condition
At goal: Percentage with most recent lab value at target (see references/panel-management.md for targets)

Diabetes: A1c < 7.0% (or < 8.0% for elderly/complex)
Hypertension: BP < 130/80 (or < 140/90 for elderly)
CKD: eGFR stable or improving

Not at goal: Percentage above target
No recent lab: Percentage with no relevant lab in past 12 months
No recent visit: Percentage with no ambulatory encounter in past 6 months
Complication rate: Percentage with documented complications (query additional Conditions)

Step 6: Risk Stratify the Cohort

Assign risk tiers based on criteria in references/panel-management.md:

High risk: Lab values severely out of range, multiple comorbidities, no recent visit, recent ED/inpatient encounter
Medium risk: Lab values moderately out of range, 1-2 comorbidities, visit within 6 months
Low risk: At goal, recent visit, stable

For high-risk identification, check for recent ED visits:

Tool: fhir_search
resourceType: "Encounter"
queryParams: "patient=[patient-id]&class=EMER&date=ge[12-months-ago]&_count=10"

Step 7: Generate Outreach Priority List

Sort patients by risk tier (high first), then by days since last visit (longest first). Present top 20 patients needing outreach with:

Name, MRN, age
Most recent lab value and date
Last visit date
Risk tier and reason
Recommended action (schedule visit, repeat labs, care management referral)

Step 8: Present Panel Summary

PATIENT PANEL OVERVIEW: [Condition Name]
=========================================
Panel Size: [N] patients
Report Date: [today]

AGGREGATE METRICS
-----------------
At Goal:           [N] ([%])
Not at Goal:       [N] ([%])
No Recent Lab:     [N] ([%]) -- lab not done in 12 months
No Recent Visit:   [N] ([%]) -- no visit in 6 months

RISK STRATIFICATION
-------------------
High Risk:    [N] ([%])
Medium Risk:  [N] ([%])
Low Risk:     [N] ([%])

DISTRIBUTION (e.g., A1c ranges for diabetes)
---------------------------------------------
< 7.0%:   [N] ([%])
7.0-8.0%: [N] ([%])
8.0-9.0%: [N] ([%])
> 9.0%:   [N] ([%])

TOP OUTREACH PRIORITIES
-----------------------
1. [Name] (MRN: [mrn]) - A1c 11.2% (6 months ago) - No visit in 9 months - ACTION: Schedule urgent visit
2. [Name] (MRN: [mrn]) - A1c 9.8% (3 months ago) - ED visit last month - ACTION: Care management referral
...

Examples

Example 1: Diabetes Panel Overview

User says: "Show me my diabetes panel. How many patients are at goal?"

Actions:

Search Condition with code 44054006 (T2DM), clinical-status active. Returns 87 patients.
Batch-fetch Patient demographics for all 87.
Search Observation for LOINC 4548-4 (A1c) in past 12 months. Match to cohort.
Search Encounter for ambulatory visits in past 6 months. Match to cohort.
Compute: 52/87 (59.8%) at A1c < 7%, 23/87 (26.4%) between 7-9%, 7/87 (8.0%) above 9%, 5/87 (5.7%) no A1c in 12 months.
Risk stratify: 12 high risk, 28 medium, 47 low.

Result:

PATIENT PANEL OVERVIEW: Type 2 Diabetes Mellitus
=================================================
Panel Size: 87 patients
Report Date: 2024-11-15

AGGREGATE METRICS
-----------------
At Goal (A1c < 7%):   52 (59.8%)
Not at Goal:           30 (34.5%)
No Recent A1c:          5 (5.7%)
No Recent Visit:        9 (10.3%)

A1c DISTRIBUTION
----------------
< 7.0%:   52 (59.8%)
7.0-8.0%: 16 (18.4%)
8.0-9.0%:  7 (8.0%)
> 9.0%:    7 (8.0%)
No data:   5 (5.7%)

TOP OUTREACH PRIORITIES
-----------------------
1. James Carter (MRN: 44521) - A1c 12.1% (4mo ago) - No visit 8 months - Schedule urgent visit + endo referral
2. Lisa Tran (MRN: 33876) - A1c 10.4% (6mo ago) - ED visit for DKA last month - Care management referral
3. Robert Kim (MRN: 55903) - No A1c in 14 months - No visit 11 months - Schedule visit + labs

Example 2: Hypertension Panel for Outreach

User says: "Which of my hypertension patients haven't been seen recently and are uncontrolled?"

Actions:

Search Condition with code 38341003 (hypertension), active. Returns 142 patients.
Search Observation LOINC 85354-9 (BP panel) in past 12 months.
Search Encounter ambulatory in past 6 months.
Filter to patients with most recent systolic >= 140 AND no visit in 6+ months.

Result:

UNCONTROLLED HYPERTENSION - NO RECENT VISIT
============================================
Found 18 of 142 hypertension patients with BP >= 140/90 and no visit in 6+ months

1. Dorothy Williams (MRN: 22814) - Last BP: 168/94 (5mo ago) - Last visit: 8 months ago
2. Frank Nguyen (MRN: 31005) - Last BP: 158/92 (7mo ago) - Last visit: 10 months ago
3. Patricia Jones (MRN: 45221) - Last BP: 152/88 (4mo ago) - Last visit: 7 months ago
...

RECOMMENDED ACTIONS:
- All 18 patients: Schedule follow-up visit within 2 weeks
- Patients with BP > 160 systolic (2): Consider same-week appointment
- Patients with no visit > 9 months (4): Flag for care management outreach call

Example 3: Multi-Condition Panel Summary

User says: "Give me a summary of all my chronic disease panels."

Actions:

Query Conditions for each supported disease (diabetes, HTN, CHF, COPD, CKD, depression) in parallel.
For each, pull aggregate labs and visit data.
Present combined summary table.

Result:

PRACTICE PANEL SUMMARY
======================
Condition           | Patients | At Goal | Not at Goal | No Recent Lab | No Recent Visit
--------------------|----------|---------|-------------|---------------|----------------
Type 2 Diabetes     |       87 |  59.8%  |     34.5%   |     5.7%      |     10.3%
Hypertension        |      142 |  68.3%  |     24.6%   |     7.0%      |     12.7%
Heart Failure       |       31 |  71.0%  |     19.4%   |     9.7%      |      6.5%
COPD                |       23 |  60.9%  |     26.1%   |    13.0%      |     17.4%
CKD                 |       19 |  52.6%  |     36.8%   |    10.5%      |     15.8%
Depression (PHQ-9)  |       54 |  44.4%  |     37.0%   |    18.5%      |     14.8%

TOTAL UNIQUE PATIENTS: 289 (some patients appear in multiple panels)
HIGHEST PRIORITY: Depression panel has lowest at-goal rate and highest no-recent-lab rate

Troubleshooting

Condition search returns zero patients despite known panel

The FHIR server may use ICD-10 codes instead of SNOMED. Retry with ICD-10: code=http://hl7.org/fhir/sid/icd-10-cm|E11 for diabetes, I10 for hypertension, I50 for CHF.
Some systems store conditions with clinical-status=recurrence or clinical-status=relapse instead of active. Broaden: clinical-status=active,recurrence,relapse.
Check if the server requires the full SNOMED URL or accepts short codes.

Lab searches return too many results or timeout

Add patient filter if querying for a small cohort: patient=[id1],[id2]&code=....
For large panels, use _summary=count first to estimate volume, then paginate with _count=50&_offset=0.
Reduce date range from 12 months to 6 months if volume is excessive.
Some servers do not support batch patient IDs in a single query. Fall back to individual queries for the top-priority patients only.

Encounter class codes vary between FHIR servers

EPIC uses class=AMB for ambulatory. Cerner may use class=outpatient. Generic servers may use class=http://terminology.hl7.org/CodeSystem/v3-ActCode|AMB.
If encounter search returns zero results, try without the class filter and inspect returned encounter classes to determine the correct code.

quality-measure-dashboard -- calculate formal HEDIS/CMS quality measures for the panel
chronic-disease-registry-query -- deeper registry-level queries with severity stratification
care-gap-identifier -- identify specific care gaps for individual patients in the panel
preventive-care-compliance-report -- preventive care audit across the panel

patient-panel-overview— agent skill

Is patient-panel-overview safe to install?

Patient Panel Overview

Overview

FHIR Resources Used

Instructions

Step 1: Identify Target Cohort

Step 2: Retrieve Patient Demographics for Cohort

Step 3: Pull Aggregate Lab Values

Step 4: Check Recent Visit Status

Step 5: Compute Panel Metrics

Step 6: Risk Stratify the Cohort

Step 7: Generate Outreach Priority List

Step 8: Present Panel Summary

Examples

Example 1: Diabetes Panel Overview

Example 2: Hypertension Panel for Outreach

Example 3: Multi-Condition Panel Summary

Troubleshooting

Condition search returns zero patients despite known panel

Lab searches return too many results or timeout

Encounter class codes vary between FHIR servers

Audit the pieces. Scan the whole. Decide.

Is patient-panel-overview safe to install?

Patient Panel Overview

Overview

FHIR Resources Used

Instructions

Step 1: Identify Target Cohort

Step 2: Retrieve Patient Demographics for Cohort

Step 3: Pull Aggregate Lab Values

Step 4: Check Recent Visit Status

Step 5: Compute Panel Metrics

Step 6: Risk Stratify the Cohort

Step 7: Generate Outreach Priority List

Step 8: Present Panel Summary

Examples

Example 1: Diabetes Panel Overview

Example 2: Hypertension Panel for Outreach

Example 3: Multi-Condition Panel Summary

Troubleshooting

Condition search returns zero patients despite known panel

Lab searches return too many results or timeout

Encounter class codes vary between FHIR servers

Related Skills

Audit the pieces. Scan the whole. Decide.