Reconcile the export set before anyone relies on it.

01 / BUILD THE SOURCE SET

Collect records from every place that held or moved value.

A clean-looking report can still be incomplete. Start with the full period from every exchange, wallet, brokerage, bank, and tax aggregation tool that participated in the activity. Include closed accounts and inactive wallets if they contain acquisitions, transfers, fees, or dispositions that explain later records. For assets carried into the year, earlier acquisition history may be needed to establish basis.

Use exports created directly by the source application. Do not rename columns or merge files unless necessary; original headers help format detection. Never upload wallet seed phrases, private keys, passwords, API secrets, identity documents, or images. This service accepts data exports, not access credentials.

02 / NORMALIZE AND SCOPE

Different labels must resolve to one review vocabulary.

The pipeline recognizes common Koinly, CoinLedger, Coinbase, Etherscan-style wallet and token exports, brokerage histories, 1099-B detail, bank statements, and a generic signed-amount layout. It normalizes recognized rows into a shared movement record so the same checks can run across sources. Format detection is based on headers and content, not only the filename.

Select a tax year to narrow dated activity, or choose all years when investigating carry-forward issues. Undated rows stay in scope so missing dates do not silently hide unresolved records. Add known client wallet addresses only as public addresses, one per line. They help distinguish movements between the client’s own accounts from transactions involving an outside counterparty.

1. Preserve source identity.Keep account, wallet, chain, and export names recognizable. The packet’s source register should map cleanly back to the original provider.
2. Check the period.Confirm that the first and last rows cover the intended date range and that the provider did not silently limit the export.
3. Check time zones.Near-midnight events can land on different dates in local time and UTC. Record the export’s time-zone convention for the preparer.
4. Retain untouched copies.Store original exports separately. The cleanup packet is a workpaper, not a replacement for source records.

03 / INTERPRET THE REVIEW QUEUE

A flag is a question to resolve, not a tax conclusion.

Some checks use time, amount, asset, account, and ownership clues to propose a match. Similar amounts close together in time can be evidence of a transfer, but they do not prove ownership or treatment. A preparer may need statements, transaction hashes, wallet explorers, trade confirmations, or client answers to resolve the item. The review workbook keeps candidate pair identifiers and source references so the decision can be documented.

04 / USE THE PACKET AS A WORKPAPER

Resolve high-impact gaps before polishing low-impact labels.

The download contains an Excel review workbook, issue CSVs, client questions, a Markdown summary, and a JSON audit manifest. The manifest records the selected year, enabled checks, source profiles, and result counts. That makes it easier to reproduce what was reviewed and to explain why a later run changed.

• Confirm that every expected source appears in the source register and that row counts are plausible.
• Start with missing basis, incomplete trade legs, large value gaps, and records that make balances impossible.
• Resolve likely duplicates and own-account transfers before interpreting gross proceeds or income totals.
• Use generated client questions as a starting point; remove irrelevant prompts and add the facts the preparer actually needs.
• Document the evidence and human decision for each material adjustment outside the automated packet when required by the preparer’s workflow.
• Re-export corrected data from the system of record when possible, then rerun the same scope and compare the audit manifest.

The readiness index is a triage heuristic, not a measure of filing accuracy, legal compliance, audit risk, tax liability, or preparer approval. A high score cannot prove that the source set is complete; it reports only on the records and checks available to the run.

05 / PRIVACY AND RETENTION

Limit the data before it enters the workflow.

Uploaded CSV and TSV files are removed after reconciliation completes successfully. The generated packet is available under a random download token for up to one hour; expired downloads are rejected, and a background cleanup job scheduled every 15 minutes removes expired job directories. Downloads use Cache-Control: no-store. Operational logs may include a job reference and error details, but the application does not intentionally log exported rows, private keys, or seed phrases.

Remove columns the review does not need, especially full names, physical addresses, tax identifiers, and unrelated account notes. Keep the mapping between any redacted identifiers and real clients inside the preparer’s approved system. Do not email exports to support. If the records are subject to professional, contractual, or regulatory handling requirements, confirm that this workflow is approved before uploading.

Tax Pre-Cleaner currently has no advertising, analytics, or third-party tracking tags. If advertising is added in the future, it will be considered only for an explicit allowlist of public guides such as this one—never intake, results, packet downloads, or generated files—and only after the privacy disclosure, consent controls, and route-specific Content Security Policy are ready.