Claude to organize inherited land preparation
Tsuyoshi Yamasaki's relatives had more than 30 scattered plots, outdated ownership records and unused buildings. He knew about Japan's inherited-land return system but did not know how to organize the preparation. Claude Code received the family's incomplete description and researched official ministry pages and legal text before building a plan.
note.com
Original prompt
Shared by the source author. See source notes for context.
私の整理した情報全体を理解し、制度について公式情報を収集してください。
Reach people exploring AI agents
See placements →Evidence at a glance
Sourced account- Source author
- Tsuyoshi Yamasaki
- Original account
- Read the source ↗
- Published
- Source link checked
- Reported result
- The account reports three Backlog documents, one parent issue and 11 linked child tasks; no land transfer or application completion is reported.
- Human involvement
- Yamasaki corrected context and checked findings; Cursor supplied a second review, and official consultation remained necessary.
A linked account supports attribution. It does not mean we independently reproduced its result.
Report an issue with this case
Open correction emailOpens your email app with the case link included. Nothing is sent automatically.
How the agent did it
- Provide the family's existing records and unresolved assumptions.
- Draft research and tasks in Markdown, checking official sources and human corrections.
- Review with a second model, then publish linked Backlog documents and parent-child tasks.
Tsuyoshi Yamasaki's relatives had more than 30 scattered plots, outdated ownership records and unused buildings. He knew about Japan's inherited-land return system but did not know how to organize the preparation. Claude Code received the family's incomplete description and researched official ministry pages and legal text before building a plan.
The agent first drafted Markdown for review. It then used the bee command-line tool and existing task-writing skills to publish Backlog documents and a parent issue with 11 child tasks. The final set contained a research memo, an overall plan and a flow for asking the municipality about records. Tasks linked back to the relevant document sections.
Yamasaki corrected a misunderstanding about land classification and added missing context about access and boundaries. A separate review through Cursor found further omissions, including the sequencing of demolition and consultation. The author checked those findings against official wording before incorporating them.
Backlog's document API could not update existing documents, so corrections required recreating them. Yamasaki concluded that the second review should happen before publication. The finished result was a reviewed preparation system, not a completed land transfer. He still planned to confirm the family's specific situation with the responsible legal affairs office before proceeding.
Prompt
Original prompt
Source: note.com · Tsuyoshi Yamasaki
Frequently asked questions
Why draft before publishing in Backlog?
The document API could create but not update documents, so later corrections required deletion and recreation.
What did the second review change?
It exposed missing deadline and access details and the risk of ordering demolition before confirming the land's eligibility.
Was the administrative process completed?
No. The output was preparation material and tasks; the author planned case-specific confirmation with the responsible office.
More details
About this account
Claude researched official material, drafted the plan and operated bee to create linked Backlog documents and issues.
Firsthand Japanese source and translation opened and read. Prompt is a verbatim short excerpt of the author's published, privacy-redacted original instruction; legal conclusions remain case-specific preparation.
Credit: Tsuyoshi Yamasaki
Source published
Added to Agents In Use
Source checked
Tools and topics
Claude Code · Backlog · bee CLI · Cursor
Inherited land · Backlog tasks · Official-source review
Editorial completeness: 100/100. Measures documented fields, not agent performance or verified outcomes.