skills/hsa-receipt-vault/SKILL.md
Maintains a ledger of HSA-qualified medical expenses paid out of pocket (not reimbursed from the HSA), each with date, amount, provider, description, and a path to the scanned receipt. Tracks the running unreimbursed total — the amount of HSA balance the household can pull tax-free at any future date — and validates each candidate against IRS-qualified-expense categories. Use when a new medical receipt arrives, when totaling future tax-free HSA reimbursements, when planning a deferred reimbursement, or when user mentions HSA receipt vault, qualified medical expenses, or HSA shoebox strategy.
npx skillsauth add lyndonkl/claude hsa-receipt-vaultInstall this skill globally with one command. Works with Claude Code, Cursor, and Windsurf.
3 of 9 scanners reported clean
Some scanners were skipped, did not run, or reported a non-clean status. Review each row below.
The HSA "shoebox" strategy: pay medical bills out of pocket, save the receipts, let the HSA grow tax-free for decades, then reimburse yourself any time later — there is no time limit on HSA reimbursements as long as the expense was incurred after the HSA was opened. This skill maintains the ledger, validates entries, and exposes the running unreimbursed total.
Every dollar in this ledger is a dollar the household can withdraw tax-free from the HSA at any future point — effectively making the HSA function like a Roth IRA for medical-expense-equivalent dollars, with the optionality to reimburse before retirement if cash is needed.
Two modes:
A — single receipt to add:
{
"date": "2026-03-10",
"amount_cents": 18500,
"provider": "Dr. Smith DDS",
"description": "Dental cleaning",
"receipt_path": "receipts/2026/2026-03-10-smith-dds.pdf",
"paid_from_account_id": "acc_chk_001",
"tx_id": "tx_20260310_004"
}
B — bulk reconcile (scan transactions for category: health.*):
{
"transactions": [...],
"existing_vault": [...],
"hsa_open_date": "2018-01-15"
}
In bulk mode, the skill walks category: health.* transactions, proposes vault entries for those not already in the ledger, and asks the user (via output) to confirm each.
IRS Publication 502 governs HSA-qualified expenses. The skill validates against these high-level categories:
| Qualified | Examples | |---|---| | Medical care | Doctor visits, hospital, surgery, ambulance | | Dental | Cleanings, fillings, extractions, orthodontia | | Vision | Eye exams, glasses, contacts, LASIK | | Mental health | Therapy, psychiatry, prescribed inpatient treatment | | Prescription drugs | Rx, insulin (OTC insulin since 2020) | | OTC medicine and menstrual products | Allowed since CARES Act 2020 | | Medical equipment | Crutches, hearing aids, CPAP, blood-pressure monitors | | Long-term care | Premiums (within IRS limits) and qualified care | | Acupuncture, chiropractic | Yes | | Smoking cessation | Yes | | Weight loss for specific diseases | Only if prescribed for diabetes, hypertension, etc. |
Generally not qualified:
If a candidate is ambiguous, mark qualification: unclear with the IRS Pub 502 reference and let the user confirm.
HSA Vault Progress:
- [ ] Step 1: Validate receipt date is after HSA open date
- [ ] Step 2: Map description to a qualified-expense category
- [ ] Step 3: Confirm receipt_path exists or note it's pending
- [ ] Step 4: Confirm amount paid from a non-HSA account (else not a vault entry)
- [ ] Step 5: Generate id and append to ledger
- [ ] Step 6: Update running unreimbursed_total
- [ ] Step 7: Emit warnings for ambiguous qualifications
Receipts dated before the HSA was opened are NOT eligible for tax-free reimbursement, no matter when the HSA grows. Reject with a clear warning.
Use the description and the merchant category to map to a qualified category. If unmappable, mark qualification: unclear and suggest the user attach a Letter of Medical Necessity if applicable.
The receipt must be retained — the IRS can request substantiation up to the statute of limitations after reimbursement. Acceptable receipt formats: PDF, JPG, PNG. Path stored relative to the household-finance root: receipts/YYYY/YYYY-MM-DD-provider.ext.
If receipt_path is missing, accept the entry but mark receipt_status: pending and emit a warning. Do not allow pending entries to count toward the running unreimbursed_total — they will, once the receipt is attached.
A receipt only enters the vault if it was paid from a non-HSA account. Anything paid directly from the HSA debit card or HSA bill-pay is already a tax-free distribution; it doesn't go in the shoebox.
Generate id: rcpt_<year>_<seq>, append to hsa.json.receipt_vault[]. Never mutate prior entries; if a correction is needed, append a correction_of: <prior_id> entry.
unreimbursed_total_cents = Σ amount_cents WHERE receipt_status = "ok" AND reimbursed = false.
This is the headline number — the dollars the household can pull tax-free at any time.
qualification: unclear — surface to user.For mode A (single add):
{
"added": {
"id": "rcpt_2026_0014",
"date": "2026-03-10",
"amount_cents": 18500,
"provider": "Dr. Smith DDS",
"description": "Dental cleaning",
"category": "dental",
"qualification": "ok",
"receipt_path": "receipts/2026/2026-03-10-smith-dds.pdf",
"receipt_status": "ok",
"paid_from_account_id": "acc_chk_001",
"tx_id": "tx_20260310_004",
"reimbursed": false,
"tax_year": 2026
},
"running_totals": {
"unreimbursed_total_cents": 4435500,
"by_year_cents": {
"2024": 1820000,
"2025": 2150000,
"2026": 465500
}
},
"warnings": []
}
For mode B (bulk reconcile):
{
"proposed": [
{
"tx_id": "tx_20260118_005",
"amount_cents": 4500,
"provider": "CVS Pharmacy",
"description_raw": "CVS/PHARMACY #04123",
"suggested_category": "prescription_drugs",
"qualification": "ok",
"needs_receipt_upload": true
},
{
"tx_id": "tx_20260201_002",
"amount_cents": 8500,
"provider": "Equinox",
"description_raw": "EQUINOX MONTHLY DUES",
"suggested_category": null,
"qualification": "not_qualified",
"rationale": "Gym memberships are not qualified unless prescribed for a specific medical condition with a Letter of Medical Necessity."
}
],
"running_totals": {
"unreimbursed_total_cents": 4435500
}
}
receipt_path faithfully; flag missing receipts; do NOT count them toward the unreimbursed total.correction_of to amend.testing
Cluster a conference's event records into a small set of coarse themes with finer sub-clusters, an explicit outlier bucket, and soft (multi-membership) affinities — using the hybrid embed-then-label pipeline (embed abstracts, reduce, density-cluster, then LLM-label the clusters) when embedding libraries are available, and an LLM-reasoned hierarchical fallback when they are not. Embeddings do the grouping; the LLM only names the groups. Conference-agnostic. Use when turning structured event records into a navigable theme map for preference elicitation and scheduling, when you need 6-8 reasonable themes rather than 20 muddy ones, or when overlapping talks must belong to more than one theme. Trigger keywords - theme clustering, cluster talks, embed then label, soft membership, outlier talks, conference themes, topic map.
development
Build a personal conference schedule as a constraint-optimization problem — hard constraints (no time overlap, room-to-room travel time, capacity/registration, the attendee's own must-attends and blackouts) plus a user-owned weighted objective trading interest against breadth, pacing (maximize contiguous free time), and serendipity. Surfaces unbreakable conflicts (two high-value overlapping talks the model cannot rank) as decisions for the human rather than silently picking, and reports what each choice traded away. Conference-agnostic. Use to turn a preference profile plus a theme map into a day-by-day plan, to resolve overlapping sessions, or to balance a packed vs paced schedule. Trigger keywords - schedule optimization, conference schedule, constraint optimization, overlapping talks, contiguous free time, conflict surfacing, packed vs paced.
development
Parse a heterogeneous conference program (markdown, HTML, PDF-derived text, or JSON) into normalized event records with per-field confidence scores and independent classification axes (topic, depth, format, prerequisites, recorded, capacity). Detects the program's format before extracting, treats every inferred field as uncertain (present vs inferred vs missing), and flags thin or missing abstracts so downstream enrichment can target them. Conference-agnostic. Use when ingesting a conference or event schedule into a structured store, normalizing a talk/session list, or extracting per-session metadata with calibrated confidence. Trigger keywords - program ingestion, parse schedule, session extraction, event records, conference program, talk metadata, per-field confidence.
development
Build a personalized preference profile from a small number of well-chosen, cluster-grounded questions instead of a long survey. Represents the person's interests as an uncertainty region over the theme map, picks the single highest-information-gain choice-based question (contrasting real talks from different clusters), balances exploiting known interests against exploring uncertain ones, deliberately injects outlier probes to fight selection bias, and stops as soon as the schedule would be stable. Also elicits the user-owned objective weights and hard constraints. Interactive — runs where it can actually ask the person. Conference-agnostic. Use to turn a theme map into a preference profile, to decide what to ask a conference attendee, or to elicit scheduling priorities. Trigger keywords - preference elicitation, ask few questions, information gain, choice-based questions, selection bias probe, objective weights, attendee preferences.