Japan · Async only
00

What to have ready before you appeal a suspended listing

Written 11 August 2026 by Desiria LLC. The tool described at the end is MIT and on GitHub, with tests you can run.

This page is about the record, not the outcome. Whether an appeal succeeds is the platform's decision and nobody outside it can promise you anything. What you can control is whether you can show, precisely, what you changed and when — and that turns out to be the part most people assemble badly, under time pressure, from screenshots scattered across a laptop.

01

Why the record is usually the weak part

The fixes happen before the record does
You find the problem, you fix it across several products, and only afterwards do you think about proving it. By then the before-state is gone — the page has already changed.
Screenshots lose their context
A folder of images with camera-roll filenames tells a reviewer nothing about which product it was, what was wrong, or when it changed. The evidence exists; the connection to the claim doesn't.
A second appeal starts from zero
If the first one doesn't land, the useful question is "what is different this time" — and that requires the first submission to still exist in a comparable form. Usually it doesn't.
02

What a usable record contains

One entry per change, with a time

What was changed, on which item, and when. Not "we updated our policies" but a dated list a reader can check line by line. The timestamp is what makes the claim verifiable rather than asserted.

Evidence tied to the entry that needs it

Each file linked to the specific change it supports. A screenshot on its own is decoration; a screenshot attached to "changed the returns policy on SKU-1001 on 3 August" is evidence.

A hash, so a later edit is detectable

Record a checksum of each file at the time you assemble the pack. A hash doesn't stop anyone editing the file — it just means that if the copy in your hands stops matching the recorded value, you can tell. That's enough to distinguish what you actually submitted from what the file became afterwards.

A diff against the previous submission

If there was an earlier attempt, the only genuinely new information is what changed since. Being able to state that plainly saves the reader work — and saves you from re-sending the same case.

03

The part that bit us: where evidence is allowed to come from

This is the bit worth stealing even if you never touch our code.

If you script the assembly of an evidence bundle, the script reads files and copies them into something you then send outside your company. That means the set of paths it will follow is a security question, not a convenience one.

Checking that a path string starts with your evidence folder is not enough. A symlink inside that folder can point anywhere, and a naive check passes it — the file gets read, hashed, and copied into the bundle. Resolve the real path and check that, and make the evidence directory mandatory rather than optional: without one there is no boundary to enforce, and a default of "wherever the file happens to be" quietly means "anywhere".

We know because our own first version did exactly that, and a pre-publication review caught it before release rather than after.

04

The tool

desiria-review-dossier takes a list of changes you wrote and a folder of evidence, and produces one document you can attach: a dated change log, an index of the evidence with checksums, and — if you point it at a previous submission — what moved since. MIT, runs locally, no account.

github.com/desiria-ja/desiria-review-dossier

What it deliberately does not do: it does not judge your case, advise you on what to fix, act on your account, or predict an outcome. It does not connect to any platform API. It organises what you tell it, and records where each file came from.

Corrections welcome — if something here is wrong, tell us at contact@byrdhq.com and we'll fix the page.

05

About this page

Desiria LLC builds data-integration and record-keeping tools. We wrote this because assembling the record is the part that gets done last and worst. Desiria LLC is an independent third party. We are not affiliated with, endorsed by, or acting on behalf of any marketplace or platform whose review processes this page describes, and we have no ability to influence their decisions. All trademarks belong to their respective owners. More about us · Privacy and subprocessors

Disclosure: this page and the linked code are the work of an AI agent operating on behalf of Desiria LLC.