Install any skill in seconds. Free to start, no credit card required.
Get Started Free →Automate payer review of prior authorization (PA) requests. This skill should be used when users say "Review this PA request", "Process prior authorization for [procedure]", "Assess medical necessity", "Generate PA decision", or when processing clinical documentation for coverage policy validation and authorization decisions.
.claude/skills/prior-auth-review-skill/SKILL.md| Model | Eval pass | Runs |
|---|---|---|
| gemini-3.6-flash | 100% | 6 |
| Test case | Without → With | Effect | Δ tokens | Δ turns |
|---|---|---|---|---|
| case-06 | ✗→✓ | ▲ Improved | — | — |
| case-15 | ✗→✓ | ▲ Improved | — | — |
| case-22 | ✗→✓ | ▲ Improved | — | — |
| case-14 | ✗→✓ | ▲ Improved | — | — |
| case-05 | ✗→✓ | ▲ Improved | — | — |
This skill automates the payer review process for prior authorization (PA) requests. It processes clinical documentation, validates medical necessity against coverage policies, and generates authorization decisions with supporting rationale.
Target Users: Health insurance payer organizations (Medicare Advantage, Commercial, Medicaid MCOs)
Value Proposition: Reduce PA review time from 30-60 minutes to under 5 minutes. Enable auto-approval for 40-60% of clear-cut cases.
This skill uses a simplified 2-subskill workflow:
Subskill 1: Intake & Assessment
↓ (validates data, extracts clinical info, assesses medical necessity)
Subskill 2: Decision & Notification
↓ (generates auth decision with provider notification)
ONLY REVIEW THE SUBSKILL FILES WHEN THEY ARE NEEDED, DONT PRE-READ THE WHOLE SKILL ON BOOTUP
Output: Authorization Decision Packagewaypoints/
├── assessment.json # Subskill 1 output (consolidated)
└── decision.json # Subskill 2 output (final decision)This skill requires 3 healthcare MCP connectors:
For detailed tool usage, parameters, and CMS web resources, see references/01-intake-assessment.md.
During execution, the skill displays notifications before and after each MCP connector call:
See README.md File Organization section for complete directory structure and file descriptions.
This skill enforces a decision policy rubric that determines the outcome when validation checks fail. The policy balances regulatory compliance, patient safety, and operational efficiency.
See references/rubric.md for:
Quick Summary:
To customize decision logic for your organization, edit references/rubric.md.
Simply invoke the skill:
Use the prior-auth-review-skillThe skill will:
When this skill is invoked:
Before proceeding, verify required MCP connectors are available.
Check for the following MCP connectors:
If any MCP connectors are not configured:
Display error and exit: > "Missing required MCP connectors: list missing connectors]. This skill requires all three healthcare MCP connectors to function. Please configure the missing connectors and try again. See README Prerequisites for setup instructions."
Exit skill.
If all MCP connectors are available: Proceed silently to next step.
Prompt the user to provide input files or use sample data.
Display the following prompt:
Prior Authorization Review requires the following input files:
REQUIRED FILES:
1. Prior Authorization Request Form (PDF) - Contains member info, requested service, provider details
2. Clinical Notes / H&P (PDF) - History and physical examination documentation
3. Diagnostic Imaging Reports (PDF) - CT, MRI, X-ray, or other imaging results
4. Laboratory Results (PDF) - Relevant lab work supporting medical necessity
5. Additional Supporting Documentation (PDF, optional) - PFTs, specialist consults, etc.
OPTIONS:
(A) Upload your own files - Provide paths to each required document
(B) Use sample files - Load pre-configured sample case (CT-guided lung biopsy)
Enter your choice (A/B): ___If user selects (A) - Upload own files:
using_sample_files = FalseIf user selects (B) - Use sample files:
assets/sample/:01_Prior_Auth_Request_Form.pdf02_Clinical_Notes_H_and_P.pdf03_CT_Chest_Report.pdf04_Laboratory_Results.pdf05_Pulmonary_Function_Tests.pdfusing_sample_files = True1234567890) and sample member ID (1EG4-TE5-MK72). This combination triggers demo mode, which skips the NPI MCP lookup for this specific provider only. All other MCP calls (ICD-10 validation, CMS Coverage policy search) execute normally.Check if waypoints/assessment.json exists:
Found incomplete PA request: [Request ID] Resume this request? (Y/N): ___
Execute: Read and follow references/01-intake-assessment.md
What it does:
Output: waypoints/assessment.json (consolidated)
Duration: 3-4 minutes
Ask user:
Ready to proceed to Subskill 2? (Y/N): ___Execute: Read and follow references/02-decision-notification.md
What it does:
Output:
waypoints/decision.json (final decision)outputs/notification_letter.txt (provider notification)Duration: 1-2 minutes
Display a concise completion message with:
Offer user options to:
Missing MCP Servers: If required MCP connectors not available, display error listing missing connectors and Removefully.
Missing Subskill Prerequisites: If Subskill 2 invoked without waypoints/assessment.json, notify user to complete Subskill 1 first.
File Write Errors: If unable to write waypoint files, display error with file path, check permissions/disk space, and offer retry.
Data Quality Issues: If clinical data extraction confidence <60%, warn user with confidence score and low-confidence areas. Offer options to: continue, request additional documentation, or abort.
For all errors, provide clear, actionable messages and user options for resolution.
Before completing workflow, verify:
waypoints/assessment.json on startup. If found and status is not "assessment_complete", offer to resume.CRITICAL: Every time you invoke an MCP tool or WebFetch for code validation:
BEFORE the call:
AFTER receiving results:
If there's an issue:
Benefits:
Requirements:
waypoints/assessment.json (consolidated)waypoints/decision.json and notification letter| Case | Status | Duration (ms) | Turns | Tokens | Tool calls | ||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
| Without | With | Δ | Without | With | Δ | Without | With | Δ | Without | With | Δ | ||
case-06 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-09 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-12 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-15 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-22 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-14 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-11 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-03 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-13 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-05 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-18 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-16 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-10 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-07 | pass→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-02 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-21 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-19 | fail→pass | — | — | — | — | — | — | — | — | — | — | — | — |
case-08 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-17 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-20 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-04 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
case-01 | fail→fail | — | — | — | — | — | — | — | — | — | — | — | — |
DecimalAI ran this skill against gemini-3.6-flash twice over the same eval suite — once with the skill loaded and once without — and compared the two runs case by case. 22 cases were attempted, and 20 counted toward the lift figure. The other 2 produced results that are not comparable between the two arms, so they are excluded from the headline rather than averaged into it. The headline lift of +36 percentage points is the difference between those two pass rates over the 20 comparable cases.
The per-case answers from this run were removed by the retention sweep, so the case table below shows the verdicts without the text either arm produced. The counts above were recorded at the time and are unaffected. Answers are now kept for 180 days.
Other measured skills in the registry, with their headline benchmark lift.