taxclaw_skill

Overview

This skill extracts, stores, and exports tax documents (W-2, all 1099 variants, K-1) using on-device AI. It is local-first: documents and extracted data remain on your machine by default and a web UI runs at http://localhost:8421 for quick review. Use the CLI for ingestion, listing, and exports in CSV or JSON formats. Cloud AI can be enabled only with explicit acknowledgment in the config.

How this skill works

You provide PDFs or images to the skill via the web UI or CLI. The local AI model inspects each page, detects form type, extracts structured fields, and saves results to a local data directory (~/.local/share/taxclaw/) with config in ~/.config/taxclaw/. Exports support wide CSV, long CSV, and JSON. Network access is limited to localhost by default; cloud mode requires explicit opt-in and a privacy acknowledgment.

When to use it

• Batch-extract fields from many W-2, 1099, or K-1 forms stored on your computer.
• Prepare tax data exports for accountants or spreadsheet workflows without sending documents off-device.
• Quickly review and correct extracted fields using the local web UI at localhost:8421.
• Archive and back up parsed tax documents and structured metadata locally.
• Convert legacy scanned tax forms into machine-readable JSON or CSV for analysis.

Best practices

• Run the included setup script and start the local server before ingesting many documents.
• Keep input files organized by filer and tax year to streamline ingestion flags (--filer, --year).
• Use the web UI to validate extracted fields before exporting to avoid downstream errors.
• Restrict processing to local-only mode unless you explicitly need cloud processing and accept the privacy warning.
• Regularly back up the local data directory (~/.local/share/taxclaw/) as part of your archive strategy.

Example use cases

• Accountant needs a machine-readable CSV of all 1099s from a client’s scanned folder for reconciliation.
• Individual wants to extract W-2 wages and tax withholding fields from multiple years for personal records.
• Bookkeeping team archives K-1s and exports JSON for import into a tax-prep or ERP system.
• Compliance team audits retained tax documents and needs a searchable local dataset without external sharing.
• Developer prototypes an automated pipeline to ingest and normalize 1099 variants into a financial database.