MCP App Store
by KeenEthics
CREparser Deal Review icon

CREparser Deal Review MCP Server for Claude and ChatGPT

By Superdocs

Overview

CREparser Deal Review reads the documents behind a commercial real estate deal: rent rolls and T12 operating statements (PDF, Excel or CSV), offering memoranda, loan agreements and leases. Every figure links to the highlighted passage or spreadsheet cells it came from. Tie each rent roll unit out against its lease, check a T12's totals and NOI month by month, test an offering memorandum's claims against the property's own documents, and abstract a loan's terms, reserves and deadlines. Approve or flag each item, then export a PDF report, a spreadsheet or the highlighted source. It also checks estoppel certificates, CAM reconciliations and tenant ledgers against the leases they rely on, abstracts appraisals and tests their valuation arithmetic, compares a letter of intent or term sheet with the signed lease, and runs consent sweeps across a portfolio: which leases need the landlord's consent for an asset sale, merger or change of control.

How to connect

  1. Claude: Open the CREparser Deal Review listing in the Claude connectors directory and choose Connect. Claude listing CREparser Deal Review exposes 32 tools here.
  2. ChatGPT: Open the CREparser Deal Review listing in the ChatGPT app directory and choose Connect. ChatGPT listing CREparser Deal Review exposes 35 tools here.

Tools

appraisal_show

ChatGPT
Display a validated appraisal check: the concluded values, the direct capitalization re-done, and every figure the appraiser used with its verdict against the property's documents, the report passage highlighted, with PDF, Excel and CSV exports. Call once with the draft_id that appraisal_validate returned with ready_to_display true.

appraisal_validate

ChatGPT
Check an appraisal review before display. Works with or without the property's documents: without them it records the values, re-does the direct capitalization and keeps the report's figures as not_checked, with inconsistencies within the report as risks. List every concluded value (values) and the direct capitalization (direct_cap), each quoted; the server re-does NOI ÷ cap rate against the stated value, allowing for the appraiser's rounding. Each claim quotes the report; supported and differs claims name a source (a saved T12, rent roll, lease or loan abstract) and a fact of it or a calculation from it. The server computes the facts, re-does calculations and requires the verdict to follow from the numbers (within 0.5% for money, 0.05 points for percentages). Validation lists each source's facts (source_facts) to cite. Returns ready_to_display and errors with paths. Each call saves the checked abstract as a new temporary draft and returns its draft_id: to fix errors, call again with draft_id and changes for only the corrected items instead of resending everything. Changes apply to a copy saved as a new draft; earlier drafts, imported documents and saved abstracts are never modified or deleted.

billing_show

ChatGPT
Display a validated billing check: each charge with what the lease sets for that period, the ledger line highlighted and the lease clause behind it, the totals billed too much and too little, with PDF, Excel and CSV exports. Call once with the draft_id that billing_validate returned with ready_to_display true.

billing_validate

ChatGPT
Check a tenant ledger, invoice or rent statement against the tenant's lease before display. Each charge quotes the ledger (or cites its cells) with billed_for, the first day of the period it bills; supported and differs charges name the saved lease abstract and the item that sets the amount (item_ids and value) or a calculation from those items. Base rent must cite the rent schedule row in force on billed_for, so an abatement or a step-up is read on the right day. The server re-does calculations, requires the verdict to follow from the figures, and totals what was billed above the lease, below it and without a basis in it. Validation lists the lease's items with their ids (source_items). Returns ready_to_display and errors with paths. Each call saves the checked abstract as a new temporary draft and returns its draft_id: to fix errors, call again with draft_id and changes for only the corrected items instead of resending everything. Changes apply to a copy saved as a new draft; earlier drafts, imported documents and saved abstracts are never modified or deleted.

cam_show

ChatGPT
Display a validated reconciliation check: each billed figure against the amount the lease allows, every expense line with its treatment, the statement highlighted and the lease clause behind each term, with PDF, Excel and CSV exports. Call once with the draft_id that cam_validate returned with ready_to_display true.

cam_validate

ChatGPT
Check a CAM or operating expense reconciliation against the tenant's lease before display. Every expense line cites the statement (quote or cells) with its building total; the lease terms (share, base_year, cap, gross_up, admin_fee) and every excluded or limited line name the saved lease abstract's fields by key, and each percentage taken from the lease must be one those fields state. The server recomputes recoverable expenses, the cap, the excess over the base year, the tenant's share, the fee, the total and the balance, and compares each billed figure (within a dollar). Returns ready_to_display, errors with paths and the recomputed figures. Each call saves the checked abstract as a new temporary draft and returns its draft_id: to fix errors, call again with draft_id and changes for only the corrected items instead of resending everything. Changes apply to a copy saved as a new draft; earlier drafts, imported documents and saved abstracts are never modified or deleted.

cre_account

ChatGPT
Show the signed-in CREparser account's plan and how many documents it has left, with a link to where its plan is described. A document is used when its first abstract or review is shown; a lease and its amendments are one. Use when the user asks about their plan, usage or document limits. One account and one allowance cover Lease Abstraction and Deal Review, in Claude and ChatGPT.

cre_combine_pdfs

ChatGPT
Combine separately imported PDFs (an agreement and its amendments, modifications or side letters, each with its own session_id) into one document, in the order given, so one abstract can cite all of them. Returns a new session_id and each file's page range in the combined document; list those as the abstract's documents (add ids and dates) and use the combined page numbers. Use when the amendments arrived as separate files.

cre_delete

ChatGPT
Delete an imported document and its saved abstracts and reviews. Use only when the user asks. Does not delete the original attachment or the conversation.

cre_export

ChatGPT
Create a download for a saved abstract: abstract_pdf (formatted report with citations), xlsx, csv, ics (upcoming dates with reminders, where the document has dates), lease_accounting (a lease's ASC 842 and IFRS 16 inputs and monthly fixed payments, as Excel) or highlighted_pdf / highlighted_pages (the source PDF with cited text highlighted).

cre_load_review

ChatGPT
Return the saved review (checks, flags, edits, notes, reviewer name, market rents) for a saved abstract.

cre_open_upload

ChatGPT
Show a file picker in which the user selects a document (PDF, .xlsx or CSV) from their device and uploads it. Use only when the user asks for it or when the attachment cannot be transferred another way. Accepts the upload_id from cre_prepare_upload to reuse its ticket.

cre_prepare_upload

ChatGPT
Create a single-use ticket to upload a document that the host holds only as a local file in a code-execution sandbox. This call does not transfer the file: POST its original bytes to the returned upload_url with the returned headers and the Content-Type for its format (raw body, no redirects). HTTP 200 returns the session_id. Never print the token.

cre_read_document

ChatGPT
Import one user-selected commercial real estate document from its host-provided HTTPS attachment URL: a PDF (lease, loan, rent roll, operating statement, offering memorandum, estoppel certificate, CAM reconciliation, tenant ledger, appraisal, letter of intent, consent sweep) or, for rent rolls, operating statements, CAM reconciliations and tenant ledgers, an Excel workbook (.xlsx) or CSV. Returns a session_id and the first pages of text; spreadsheet text gives every value with its cell address (G14=60175) so abstracts can cite cells. Use when the user asks to abstract, check or analyse one of these documents.

cre_read_pages

ChatGPT
Return the extracted text of a page range from an imported document (session_id, start_page, count). Spreadsheets are split into pages of rows. Results stop at a character budget; continue from next_page.

cre_refresh_access

ChatGPT
Renew the viewer's short-lived access to the document.

cre_save_review

ChatGPT
Save the reviewer's checks, flags, edited values, notes and name for a saved abstract, and for a rent roll any market rents for vacant units.

cre_search_text

ChatGPT
Find headings, defined terms, line items or phrases in an imported document. Returns the page and surrounding text of each match, for up to 10 queries at once.

estoppel_show

ChatGPT
Display a validated estoppel check: every statement the tenant certifies with its result, the certificate passage highlighted and the lease or rent roll figure it was compared with, with PDF, Excel and CSV exports. Call once with the draft_id that estoppel_validate returned with ready_to_display true.

estoppel_validate

ChatGPT
Check an estoppel certificate review before display. Each statement quotes the certificate; supported and differs statements name a source (the tenant's saved lease abstract, or the rent roll) and a fact of it, a value its items hold (item_ids) or a calculation from them. The lease's rent is read on the certificate's date (certified_on), and dates compare exactly (unit date, YYYY-MM-DD). The server computes the facts, re-does calculations and requires the verdict to follow from the figures. Validation lists each source's facts (source_facts) and the lease's items and the rent roll's units with their ids (source_items); a rent roll's facts are property totals, so compare one unit by its item id and value. Returns ready_to_display and errors with paths. Each call saves the checked abstract as a new temporary draft and returns its draft_id: to fix errors, call again with draft_id and changes for only the corrected items instead of resending everything. Changes apply to a copy saved as a new draft; earlier drafts, imported documents and saved abstracts are never modified or deleted.

lease_save_template

ChatGPT
Save a custom abstract template for Excel and CSV exports: the user's own field list, order and labels (for example their company's abstract form or a lease administration import layout). Use when the user provides or describes their template. Map each of their fields to a catalog key (or a custom key the abstract uses); saving again with the same name replaces it.

lease_show_abstract

ChatGPT
Display a validated lease abstract in an interactive viewer: key terms, rent schedule, critical-date timeline, risk flags and every field linked to its highlighted clause in the PDF, with export to a PDF abstract, CSV and highlighted lease. Call once with the draft_id that lease_validate_abstract returned with ready_to_display true. Several leases in one call open as a portfolio. Rejects unvalidated input without saving.

lease_validate_abstract

ChatGPT
Check a complete lease abstract before display, without opening the viewer. Locates every quotation in the PDF, checks that figures in found values appear in their quotations, that calculated items show calculations, and that every core field for the property type is addressed. Validate one lease at a time, correcting all errors before moving on. Returns ready_to_display, per-lease error counts and paginated errors with paths; use draft_id and error_offset=next_error_offset to read further errors. Combine finished drafts using draft_ids without resending the abstracts. Standard field keys ( = core, required for every lease): parties: landlord, tenant, guarantor, property, premises, rentable_area, pro_rata_share, lease_date, lease_type, amendments; term: term_length, commencement_date, rent_commencement_date, expiration_date, early_access, delivery_condition; rent: base_rent, rent_escalations, rent_abatement, additional_rent, percentage_rent, sales_reporting, marketing_fund, ground_rent_reset, rent_payment, late_charge; expenses: operating_expenses, base_year, expense_cap, real_estate_taxes, utilities, management_fee, admin_fee, gross_up, expense_exclusions, audit_rights; security: security_deposit, letter_of_credit, guaranty; options: renewal_option, termination_option, expansion_rights, relocation_right, purchase_option, contraction_option, co_tenancy, kick_out; use: permitted_use, exclusive_use, prohibited_uses, operating_covenant, radius_restriction, parking, signage, building_services, loading_docks, outdoor_storage; obligations: landlord_maintenance, tenant_maintenance, roof_structure, hvac_maintenance, alterations, ti_allowance, insurance_requirements, surrender, casualty, condemnation, improvements_ownership; transfer: assignment_subletting, change_of_control, recapture, transfer_profits; legal: default_cure, holdover, subordination, estoppel, indemnification, limitation_of_liability, hazardous_materials, leasehold_mortgage, governing_law, notices*, brokers. Also core by property_type (retail: percentage_rent, sales_reporting, co_tenancy, kick_out, exclusive_use, operating_covenant, radius_restriction; ground: ground_rent_reset, improvements_ownership, leasehold_mortgage; industrial: roof_structure, hvac_maintenance, hazardous_materials). Custom snake_case keys are allowed with a label and section. Matching confirms where text appears, not legal interpretation. Each call saves the checked abstract as a new temporary draft and returns its draft_id: to fix errors, call again with draft_id and changes for only the corrected items instead of resending everything. Changes apply to a copy saved as a new draft; earlier drafts, imported documents and saved abstracts are never modified or deleted.

loan_show_abstract

ChatGPT
Display a validated loan abstract in an interactive viewer: key terms, loan timeline (closing, reporting, extension notice windows, prepayment, maturity), covenants, reserves, risk flags and every field linked to its highlighted clause, with export to a PDF abstract, Excel, CSV, calendar and highlighted PDF. Call once with the draft_id that loan_validate_abstract returned with ready_to_display true.

loan_validate_abstract

ChatGPT
Check a complete loan agreement abstract before display, without opening the viewer. Locates every quotation in the PDF, checks that figures in found values appear in their quotations, that calculated items show calculations, and that every core field for the loan type is addressed; re-does debt service, DSCR and debt yield and compares the maturity and loan dates with the critical dates. Returns ready_to_display and errors with paths to correct. Standard field keys ( = core): parties: borrower, lender, guarantor, property, loan_date, servicer, loan_documents; economics: loan_amount, future_funding, interest_rate, rate_index, spread, default_rate, interest_calculation, amortization, payment_terms, late_charge, fees; term: maturity_date, extension_options, prepayment, defeasance, release_provisions; security: collateral, reserves, cash_management, rate_cap, completion_guaranty; covenants: dscr_covenant, debt_yield_covenant, ltv_covenant, cash_trap, financial_reporting, leasing_covenant, insurance_requirements, single_purpose; metrics: underwritten_noi, annual_debt_service, underwritten_dscr, debt_yield, ltv; transfer: transfer_restrictions, permitted_transfers, assumption, other_debt; default: events_of_default, cure_periods, recourse, nonrecourse_carveouts; other: governing_law*, notices. Also core by loan_type (floating: rate_index, spread, rate_cap; construction: future_funding, completion_guaranty). Custom snake_case keys are allowed with a label and section. Each call saves the checked abstract as a new temporary draft and returns its draft_id: to fix errors, call again with draft_id and changes for only the corrected items instead of resending everything. Changes apply to a copy saved as a new draft; earlier drafts, imported documents and saved abstracts are never modified or deleted.

loi_show

ChatGPT
Display a validated lease deviation report: every term of the LOI with the lease's term beside it, changed and missing terms first, the LOI passage and the lease clause highlighted, and the terms the lease adds, with PDF, Excel and CSV exports. Call once with the draft_id that loi_validate returned with ready_to_display true.

loi_validate

ChatGPT
Check a lease deviation report before display: each term a letter of intent or term sheet sets, quoted from it and compared with the saved abstract of the lease drafted or signed from it. Supported and differs terms name the lease abstract and a fact of it, the items that hold the term (item_ids, with value for a figure) or a calculation from them; dates compare exactly. Risks may rest on lease fields alone (lease_fields) for terms the lease adds that the LOI never mentioned. The server computes the facts, re-does calculations and requires the verdict to follow from the figures. Validation lists the lease's facts (source_facts) and items with their ids (source_items). Returns ready_to_display and errors with paths. Each call saves the checked abstract as a new temporary draft and returns its draft_id: to fix errors, call again with draft_id and changes for only the corrected items instead of resending everything. Changes apply to a copy saved as a new draft; earlier drafts, imported documents and saved abstracts are never modified or deleted.

om_show

ChatGPT
Display a validated offering memorandum review: every claim with its verdict, the OM passage highlighted and the source figure it was checked against, with PDF, Excel and CSV exports. Call once with the draft_id that om_validate returned with ready_to_display true.

om_validate

ChatGPT
Check an offering memorandum review before display. Each claim quotes the OM; supported and differs claims name a source (a saved T12, rent roll, lease or loan abstract) and a fact of it or a calculation from it. The server computes the facts, re-does calculations and requires the verdict to follow from the numbers (within 0.5% for money, 0.05 points for percentages; a rounded OM figure agrees with the exact one it rounds). A comparable value is the documents' own figure, worked out without the OM's figure. Mark pro forma, Year 1, budget and stabilized figures projected: true. A figure quoted from a row of figures (several periods or scenarios) names its column heading as printed, and must sit under it; the asking price must be printed in the OM. Validation lists each source's facts (source_facts) to cite. Returns ready_to_display and errors with paths. Each call saves the checked abstract as a new temporary draft and returns its draft_id: to fix errors, call again with draft_id and changes for only the corrected items instead of resending everything. Changes apply to a copy saved as a new draft; earlier drafts, imported documents and saved abstracts are never modified or deleted.

rentroll_show

ChatGPT
Display a validated rent roll in an interactive viewer: occupancy, rent, WALT, expirations by year, each unit linked to its highlighted cells, and the tie-out against the leases, with PDF, Excel and CSV exports. Call once with the draft_id that rentroll_validate returned with ready_to_display true.

rentroll_validate

ChatGPT
Check a rent roll abstract before display. Every unit cites its row (cells such as A14:I14 in a spreadsheet, or the quoted row in a PDF) and its figures, dates and tenant name must be in those cells, each figure from its own column (in the same column order as the other rows); a stated rent roll date must be printed in the document; the totals row is checked against the units; occupied units linked to saved lease abstracts are tied out on area, rent in effect on the rent roll date, lease end, deposit and tenant name. Returns ready_to_display, errors with paths, and the checks that failed. Each call saves the checked abstract as a new temporary draft and returns its draft_id: to fix errors, call again with draft_id and changes for only the corrected items instead of resending everything. Changes apply to a copy saved as a new draft; earlier drafts, imported documents and saved abstracts are never modified or deleted.

t12_show

ChatGPT
Display a validated operating statement in an interactive viewer: NOI, EGI, expense ratio, the statement by category with each line linked to its highlighted cells, checks and flagged months, with PDF, Excel and CSV exports. Call once with the draft_id that t12_validate returned with ready_to_display true.

t12_validate

ChatGPT
Check a T12 or other monthly operating statement before display. Every line maps to a standard category (income: gross_potential_rent, vacancy_loss, concessions, bad_debt, other_income, reimbursements; expense: real_estate_taxes, insurance, utilities, repairs_maintenance, payroll, management_fee, administrative, marketing, contract_services, turnover, other_expense; subtotal: effective_gross_income, total_operating_expenses, net_operating_income; below_noi: capital_expenditures, debt_service, reserves, other_below_noi) and cites its whole row; every value must be in its own month's column of the cited row, with the sign the row prints. Re-adds each row total and subtotal, recomputes NOI per month, and flags months far from the rest of the year. Returns ready_to_display, errors with paths, facts (NOI, EGI, expense ratio) and failed checks. Each call saves the checked abstract as a new temporary draft and returns its draft_id: to fix errors, call again with draft_id and changes for only the corrected items instead of resending everything. Changes apply to a copy saved as a new draft; earlier drafts, imported documents and saved abstracts are never modified or deleted.

appraisal_show

Claude

appraisal_validate

Claude

billing_show

Claude

billing_validate

Claude

cam_show

Claude

cam_validate

Claude

cre_account

Claude

cre_combine_pdfs

Claude

cre_delete

Claude

cre_open_upload

Claude

cre_prepare_upload

Claude

cre_read_document

Claude

cre_read_pages

Claude

cre_refresh_access

Claude

cre_search_text

Claude

estoppel_show

Claude

estoppel_validate

Claude

lease_save_template

Claude

lease_show_abstract

Claude

lease_validate_abstract

Claude

loan_show_abstract

Claude

loan_validate_abstract

Claude

loi_show

Claude

loi_validate

Claude

om_show

Claude

om_validate

Claude

rentroll_show

Claude

rentroll_validate

Claude

t12_show

Claude

t12_validate

Claude

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/creparser-deal-review"><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/creparser-deal-review)