MCP App Store
by KeenEthics
Contezza AI for Financial Services icon

Contezza AI for Financial Services MCP Server for Claude

By Contezza AI

Overview

Contezza AI for Financial Services makes lending and regulatory knowledge trustworthy for AI users and agents. Nine deterministic, read-only tools answer from curated facts, not from the live web: each fact is selected from public US federal sources, checked against the page that published it, and served with its citation, effective date and status. To use it: set the tools to "Always allow", then start a question with "Using Contezza," or switch web search off, when you want the cited fact. This connector gives Claude: • Loan application screening, readiness and analysis: a mortgage pre-submission worksheet and a commercial loan spread from numbers only (ratios, reserves, DSCR, stress, covenants), screened against cited thresholds, plus an HMDA benchmark against reported outcomes. Aggregates only, historical, never a decision. • Policy and document checks: extract lending facts and regulatory references from a pasted loan document or policy, compare two texts for conflicts, and screen a policy against the cited regulatory values, flagging retired thresholds. • Cited lending rules: the CFPB ability-to-repay and qualified mortgage rules, Regulation B, Regulation Z and TRID, HMDA and capital rules; replaced rules come back as dated history, never as current. • Model governance and risk management: a checklist of actions grounded in SR 26-2, which supersedes SR 11-7, each citing the guidance sentence it comes from, with a labeled derived area for AI systems. Every answer states that it is not legal, compliance or credit advice; the worksheets are not a credit decision or a substitute for an underwriting system. The tools process only what you send and retain none of it. Demonstration or sample content is labeled as such. Production use over your own policies and loan files is a paid service, hosted or in your cloud: hello@contezza.ai. Built by Contezza AI, a product of TechMatrix AI Systems LLC. Documentation: contezza.ai/financial-services.html

How to connect

  1. Claude: Open the Contezza AI for Financial Services listing in the Claude connectors directory and choose Connect. Claude listing Contezza AI for Financial Services exposes 9 tools here.

Tools

analyze_rent_roll

Claude
Builds NOI from a rent roll of units (square feet, monthly rent, lease dates; tenants anonymized as T1..Tn, no names accepted) with caller-stated vacancy, management fee, operating expenses and reserves; returns occupancy, gross potential and effective gross income, weighted average lease term, lease rollover ladder by year, top-tenant concentration, break-even occupancy and the implied value at a stated cap rate. Deterministic; stores nothing.

benchmark_mortgage_application

Claude
Returns observed outcomes for a DTI band x combined-LTV band x purpose (optionally state and loan type) from the CFPB/FFIEC HMDA public loan-level data for the latest activity year: approval share among decisioned applications, top denial reasons, interest-rate and rate-spread quartiles, loan-amount quartiles, lender count, with the dataset, year and population stated. Aggregates only (30-application floor); historical; not a prediction or a credit decision.

check_lending_policy_consistency

Claude
Deterministically screens two pasted lending texts (or two fact packets) for conflicting figures - ratios, scores, rates, fees, amounts, terms, notice windows, capital ratios - and returns a verdict with the mismatched values and excerpts. Processes only the pasted text; stores nothing.

compute_commercial_worksheet

Claude
Computes a commercial credit spread from NUMBERS-ONLY financial statements and loan terms (no entity names or identifiers accepted): EBITDA build with stated add-backs, cash flow available for debt service, DSCR and global DSCR with guarantors, LTV / loan-to-cost, debt yield, total debt to EBITDA, interest coverage, current ratio, debt to tangible net worth, multi-period trend; a rate-shock (and CRE vacancy-shock) stress case; the caller's own covenant tests; and cited screens against public facts (SBA 7(a)/504 limits and guaranty, appraisal thresholds, legal lending limit, CRE concentration screens, leveraged-lending 6x, Reg B business notification, TRID applicability, flood). Analysis with assumptions stated; not a credit decision. Stores nothing.

compute_prequal_worksheet

Claude
Computes a broker's pre-submission readiness worksheet from a NUMBERS-ONLY borrower profile (income lines, debts, assets, housing payment or rate, loan, credit score; no names, identifiers or protected characteristics accepted): front- and back-end DTI, LTV/CLTV, reserves, cash to close; screens against cited Fannie Mae, FHA, FHFA and QM thresholds; red flags (large deposits, variable-income history); package-hygiene checks; the TRID six-item test and Loan Estimate due date; optional what-if adjustments with before/after deltas; a plain-language client summary. Not a credit decision or a substitute for DU/LPA/TOTAL. Stores nothing.

extract_lending_facts

Claude
Deterministically extracts numeric lending facts (DTI, LTV, credit-score floors, APR/APOR spreads, fees, amounts, terms, reserves, notice windows, capital ratios) and regulatory references (CFR, U.S.C., SR letters, regulation nicknames, superseded ones flagged) from pasted text into a fact packet reusable by the other tools. Up to 5 documents of 200,000 characters each per call. Processes only the pasted text; stores nothing.

get_model_risk_checklist

Claude
Returns actions grounded in the current interagency model risk management guidance (SR 26-2, 2026, which supersedes SR 11-7), each citing the guidance sentence it comes from, across development, use, validation, governance, inventory, documentation and third-party areas, plus a clearly labeled derived-guidance area for AI and LLM systems. Optionally filtered to one area.

lookup_regulatory_facts

Claude
Returns anchor-verified facts from public US federal regulatory sources (CFPB ATR/QM, Regulation B, Regulation Z/TRID, rescission, HPML, FCRA, HMDA, Regulation Q capital, PCA, SR 26-2 model risk guidance), each with its value, effective date, status and a citation to the regulator's page. Superseded rules are served as dated history. With no arguments, returns the coverage index.

screen_policy_against_regulation

Claude
Extracts thresholds from a pasted policy text (or a fact packet) and compares each to the regulatory value from the cited public source - QM APR spread, points and fees, term, Regulation B notice windows, TRID timing, rescission, HPML escrow, FCRA, capital minimums - returning within / exceeds / informational per threshold with the citation and effective date. Retired thresholds still in a policy are flagged. Processes only the pasted text; stores nothing.

Similar Apps

Listed on MCP App Store

Listed on MCP App Store

Add this badge to your site or README. It links to this page.

HTML

<a href="https://mcp-app-store.keenethics-labs.com/app/contezza-ai-for-financial-services"><img src="https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg" alt="Listed on MCP App Store" width="220" height="44"></a>

Markdown

[![Listed on MCP App Store](https://mcp-app-store.keenethics-labs.com/badges/listed-on-mcp-app-store.svg)](https://mcp-app-store.keenethics-labs.com/app/contezza-ai-for-financial-services)