Launched on OpenAI Plugins Directory—OSS 'revenue-kun' v0.5.2 for Real Estate AI Agent, Now Compatible with Claude Code and Codex

revenue-kun converts rent roll PDFs/CSVs into Excel files usable for income projections. It supports launching from Codex/Claude Code and can be used locally via a web UI. Numbers shown in images are illustrative synthetic data.

【Updated July 15, 2026】Launched on OpenAI Plugins Directory

The real estate income projection OSS 'revenue-kun' v0.5.2 has been published on the OpenAI Plugins Directory.

You can search for 'revenue-kun' in ChatGPT's plugin interface and add it.

In addition to previous access via Claude Code and Codex, general users can now use revenue-kun directly from ChatGPT.

OpenAI Plugins Directory: https://chatgpt.com/plugins?q=revenue

Primary technical information: https://github.com/signal-yield/revenue-kun/releases/tag/v0.5.2

You can search for 'revenue-kun' in ChatGPT's plugin interface and add it.

Signal Yield Advisory has released the latest version v0.5.2 of 'revenue-kun (Revenue-Kun),' an open-source software that generates income projection Excel files using the direct capitalization method from rent roll CSVs or text-extractable PDFs.

revenue-kun's Local Web UI. Upload a rent roll PDF or CSV, preview the extraction results, and generate a 3-sheet Excel file including OER and detailed cost versions.

revenue-kun's role is limited to reading rent rolls and generating three sheets: `Direct Cap_OER`, `Direct Cap_Detailed Costs`, and `Parsed Rent Roll`. It does not prompt users to select calculation methods (OER vs. detailed cost) or assumptions within the app. Instead, users input required values into the generated Excel file. revenue-kun is open-source software released under the Apache License 2.0, with all source code publicly available.

Development Background

Transferring numbers from rent rolls to income projection sheets involves repetitive manual input. Errors often occur due to challenges in distinguishing monetary vs. non-monetary entries, monthly vs. annual amounts, handling vacant units, and missing cells.

Additionally, awareness of operating expenses varies by property. Some cases have detailed expense breakdowns, while others prefer simplified estimation using a general Operating Expense Ratio (OER). To accommodate both scenarios, revenue-kun simultaneously outputs both OER-based and detailed cost-based sheets, allowing users to choose based on their situation.

revenue-kun aims to assist with these initial data transfer and aggregation tasks, outputting results into an Excel file where humans can review the content. It is not intended to fully automate operations, but rather serves as a support tool to organize preliminary calculations before expert review.

What's New in v0.5.2

- Upload CSV or text-extractable PDFs directly from the browser

- Preview number of extracted units, occupancy/vacancy status, and missing data before Excel generation

- Identify recurring ancillary incomes listed in rent rolls (e.g., water fees, parking fees, other regular income)

- Generate direct capitalization Excel file (`direct_cap.xlsx`)

- Download (simultaneous output of OER and detailed cost versions) - Calculations proceed from GPI (Gross Potential Income) to EGI (Effective Gross Income)

- Then to NOI (Net Operating Income)

Usage of OER and Detailed Cost Sheets

Use the `Direct Cap_OER` sheet when detailed expense information is unavailable.

The OER version for cases where expense details are unknown. After exporting to Excel, input the assumed OER and capitalization rate to calculate NOI and projected income value. The screen shows an example using synthetic data.

Input vacancy loss rate, bad debt rate, assumed OER (Operating Expenses ÷ EGI), capital expenditures, and capitalization rate to automatically calculate from EGI to NOI and projected income value.

When expense details are known, input individual costs (management fees, repair fees, insurance, property taxes, utilities, other operating expenses) into the `Direct Cap_Detailed Costs` sheet. You don't need all items—just input known values to automatically calculate from EGI to NOI and projected income value.

The detailed cost version for cases with known individual expenses. Input management fees, repair fees, taxes, utilities, etc., and calculate NOI and projected income value independently from the OER sheet. The screen shows an example using synthetic data.

Each sheet independently calculates NOI and projected income value. Inputs or results from one sheet do not affect the other. The choice of which sheet to use—or both—is up to the user receiving the Excel file.

Safety-Oriented Design

Recurring ancillary incomes (water fees, parking, other income) are automatically reflected into the total income (GPI) of both calculation sheets. - Never fills in missing items by guessing

- Does not treat summary rows (e.g., monthly/annual totals) as individual units These calculation processes can be verified cell-by-cell in Excel, and extraction results can be previewed before generation. However, it does not guarantee absolute accuracy or complete safety.

Recurring ancillary incomes such as water fees, parking, and other regular income are automatically reflected into the total income (GPI) of both calculation sheets without any selection steps.

Excel Output

The generated Excel file consists of three sheets: Direct Cap_OER, Direct Cap_Detailed Costs, Parsed Rent Roll From the parsed rent roll details and aggregated monthly/annual income, users can verify the calculation process in Excel—from GPI, EGI, NOI, net income for capitalization, to the projected income value using the direct capitalization method. The output amount is the 'projected income value,' not the 'income value'.

Verification

Both Local Web UI and CLI support Docker. We have confirmed Docker-based Web UI build/launch and preview/Excel generation via `/api/preview` and `/api/generate`. All 401 existing test cases (pytest) passed successfully. We also evaluated the current logic using three confidential real-world rent roll PDFs. Extraction and Excel generation were confirmed via both CLI and Local Web UI, with unit counts, rent, and common service fee aggregations matching the original documents. Automatic reflection of recurring ancillary incomes was also verified.

We evaluated the current logic using three confidential real-world rent roll PDFs. Extraction and Excel generation were confirmed via both CLI and Local Web UI, with unit counts, rent, and common service fee aggregations matching the original documents. Automatic reflection of recurring ancillary incomes was also verified.