Sokage-AI

Owner Statement Agent — Testing skill for Claude Code

Testing community

Free AI agent that turns a month-end property management export into a plain-language owner update.

How to install Owner Statement Agent

This entry records only its repository, not the path inside it, so there is no exact command to give. Open Sokage-AI/owner-statement-agent and copy the folder into ~/.claude/skills/, or the file into ~/.claude/agents/.

What Owner Statement Agent does

Free AI agent that turns a month-end property management export into a plain-language owner update. Explains every line, stops when the numbers do not tie out, never sends anything. 159 test runs, all passed.

Alternatives in Testing

M Branch Prune Lite — Git Branch Management and Test/Merge Workflow List (no test version) 139 ★
Claude Android Ninja — Agent Skill for Android development with Kotlin and Jetpack Compose, covering modular architecture, Navigation 112 ★
PM Skills — 24 product management skills across the Triple Diamond lifecycle with agentskills.io spec compliance, template 84 ★

README

Owner Statement Agent

Your accounting software produces the statement. It does not explain the month to the owner. This agent turns one closed owner export into a plain-language draft, checks the lines against the stated balance, and stops if the numbers do not tie.

It drafts. A person reviews and sends. It never touches money.

[![License: MIT](https://img.shields.io/badge/License-MIT-EE6A1A.svg)](LICENSE) [![Tests](https://img.shields.io/badge/tests-159%20runs%2C%20all%20passed-4C6A1E.svg)](test-results/) [![Models](https://img.shields.io/badge/tested-Claude%20Opus%205%20%7C%20Sonnet%205%20%7C%20Haiku%204.5-1B1715.svg)](FAILURES.md)

Start here

Start with Claude if you want the path used for the published tests. The same package also includes setup paths for ChatGPT Projects and Codex. Those two paths are packaged for use, but the published 159-run verification covers Claude only.

I want to	Go to
Set it up	Claude, ChatGPT, and Codex setup
See a safe first run	Quick test after setup
Read the agent instructions	PROMPT.md
See what it must never do	NEVER.md
Check the testing	test-results/
See known limits	FAILURES.md

You need an account with the AI tool you choose. You do not need an API key, software integration, or developer setup.

Quick test after setup

The prompt already contains the fictional Marcus Webb settings used by the three sample files. You do not need to edit those settings for this test. Follow the [Claude quick test](SETUP.md#claude-quick-test), then use three fresh chats and attach one sample to each.

Send the same line every time: