Underpayment Audit

What it does

Underpayment Audit compares what commercial payers paid (remit lines) with what your contract says they should pay (a rate table), for community hospitals and surgery centers. Send up to 200 contract rates (CPT or HCPCS code, optional modifier, rate per unit) and up to 200 remit lines (your own row label, code, units, amount paid, optional patient share or the payer's allowed amount). For every line it works out the expected amount, the amount actually allowed and the difference, in whole cents, and says which lines are underpaid, overpaid, unpaid or have no rate in your table. Totals and a by-code summary come with it. Every answer names its sources (the two lists you sent, with the contract name and effective date you gave, which are echoed back and covered by a fingerprint in the signed record) and the time it was run, and carries a signed record anyone can check against the public key below. It is arithmetic only: no model, no outside lookup, nothing stored.

Who it is for

Agents and billing teams at community hospitals, surgery centers and the revenue-cycle firms that work for them, who hold the contract rates and the remits and want a fast, checkable first pass.

Price

Free, no sign-up: 120 audits an hour per address. Paid calls are planned and not charging.

Limits

Up to 200 rates and 200 remit lines per call, 60,000 bytes in all; larger requests are refused, never cut; send big batches in parts. Contract rates are private, so none are looked up: the only sources are the two lists you send, and public fee schedules are not fetched. Money is added in exact decimals and rounded to whole cents, half up. A rate with no modifier is the default for its code. Lines whose code is not in your table are listed, not guessed. A line with nothing paid is listed as unpaid (it may be a denial) and is not counted as an underpayment. Differences under the tolerance (default 1.00 dollar a line, at most 1,000) are ignored. It does NOT read contracts or payer policies, check or assign codes, apply bundling, multiple-procedure, timely-filing, lesser-of or carve-out rules, or know why a payer adjusted a line. It is not billing, coding, legal or appeal advice, and a difference is a lead to check against the contract, not proof a payer is wrong. Do not send patient names, birth dates or member, account or record numbers: only your own row numbers. Fields it does not know are refused, and row labels or names with six or more digits are refused, but it cannot spot every identifier, so keep them out. Nothing you send is stored. The signed record covers fingerprints of the inputs, the counts and the totals, not the per-line list, which you can recompute from the stated rule.

For developers: try it

POST https://llpro.ai/mcp
{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{"name":"underpayment_audit_compare","arguments":{"contract_label":"Example PPO 2026","contract_effective_date":"2026-01-01","contract_rates":[{"code":"99213","rate":"120.00"}],"remit_lines":[{"line_id":"row1","code":"99213","units":1,"paid":"95.00","patient_responsibility":"10.00"}]}}}
-> per-line expected vs allowed, shortfall in cents, totals, sources with dates, signed record

For agents

Tool underpayment_audit_compare on https://llpro.ai/mcp and in the tools list. Signed answers are checked with the public key below (signer llpro1:3elgnreutzhlkouk4u5lxyto6e): an Ed25519 signature over the observation as canonical JSON (sorted keys, no spaces).

Public key (base64): Kp0Coo+Gouhoxjxet/GMoNgsaE30l/nDtjIFCgNhea0=

This tool exists to be found by agents that search MCP registries for payer-remit and revenue-cycle checks (road: MCP registries). It makes one signed, checkable comparison per call; keep the signed record with your work so a person can see what was compared and when.

For agents: this page as Markdown · JSON · llms.txt · MCP endpoint llpro.ai/mcp · agent card