Cedar Ridge Intake Coordination
Overview
This skill covers batch healthcare intake workflows against the Cedar Ridge Intake Coordination Portal REST API. Every task follows the same pattern: fetch data from multiple endpoints, reconcile records, apply deterministic business rules, and produce a JSON answer that conforms to a provided answer template.
Core Workflow
Follow these steps in order for every intake task:
1. Orient from the answer template
Read the answer template first. It defines:
- The required top-level keys and their types
- Every allowed enum value for every field
- Ordering rules for lists (usually ascending by ID)
- Which fields treat arrays as unordered sets
Never use a value outside the template's allowed enums. Never guess ordering.
2. Fetch all relevant data
Use the portal base URL from the task prompt. Hit the endpoints described in API Endpoints. Common patterns:
- For roster tasks: fetch patients, their charts, insurance/referral data, pharmacy lookups, and ICD codes
- For referral audits: fetch all referrals in the batch, then each referral's patient, chart, documents, and ICD codes
- For transfer reviews: fetch transfers, patients, documents, and capacity data
- For enrollment panels: fetch program candidates, then each candidate's chart and clinical history
Fetch in parallel where endpoints are independent. Use POST /query for complex multi-table reconciliations.
3. Reconcile records
Cross-reference entities by their IDs. Common reconciliation patterns are in Business Rules. Key checks:
- Compare referral ICD chapters against expected chapters for the service line
- Detect duplicate referrals for the same patient
- Flag shared insurance IDs across different patients
- Check document completeness against required document lists
- Check document freshness against staleness limits
- Verify authorization status
- Assess risk factors from chart and patient data
4. Apply business rules
Use the rules in Business Rules to determine:
- Registration/intake status for each entity
- Blocker and reason codes
- Priority tiers and follow-up cadences
- Next-contact owner and route
- Chart activation needs
Every rule is deterministic: given the same inputs, the same outputs must result.
5. Build and validate the output
Construct the JSON answer following Output Conventions. Validate:
- Every list is sorted as the template requires
- Every enum value is from the template's allowed set
- Unordered-set arrays have consistent but arbitrary ordering
- All counts in summaries match the per-entity results exactly
- No prose outside the JSON
Key Principles
- Template-driven: The answer template is authoritative for output shape and allowed values. Defer to it always.
- Deterministic rules: Business logic is rule-based, not judgment-based. Same inputs always produce same outputs.
- Ascending ID order: Unless the template says otherwise, sort lists ascending by their primary ID field.
- Unordered sets: Reason codes, issue codes, and blocker codes are sets. Order is not meaningful, but be consistent.
- Count accuracy: Every count in a summary must equal the number of entities in the corresponding state.
- JSON only: Final output must be pure JSON. No explanatory prose, no markdown wrapping.