An AI-powered document extraction skill to convert W-2s, 1099s, 1040s, and scanned tax forms into clean, typed JSON with per-field confidence flags.
Install & Download
1. ClawHub CLI
The fastest way to install a skill directly from the registry.
npx clawhub@latest install deepread-tax-forms
2. Manual Installation
Copy the skill folder to one of these locations
~/.openclaw/skills/ <project>/skills/ Priority: Workspace > Local > Bundled
3. Prompt Installation
Copy this prompt to OpenClaw to install it automatically.
Help me install deepread-tax-forms using Clawhub. If Clawhub is not installed, install it first (npm i -g clawhub).
Prefer to download?
Get the raw skill files in a ZIP archive.
What is DeepRead Tax Forms?
DeepRead Tax Forms is a specialized automation skill designed to parse tax documents like W-2s, 1099-NEC/MISC/INT/DIV, and 1040s into structured JSON format. Built for developers looking to automate financial workflows, it leverages advanced OCR and layout-aware processing to extract employer/payer details, recipient info, and exact box-by-box values from both digital PDFs and physical scans. By utilizing this integration within the ecosystem of Openclaw Skills, you can seamlessly connect document extraction with downstream financial agent actions.
Every extracted field includes a dedicated needs_review flag, allowing you to build reliable human-in-the-loop validation flows. This prevents invalid, low-confidence data from entering tax filings or underwriting systems, making it one of the most reliable Openclaw Skills for fintech and accounting automation.
DeepRead Tax Forms Use Cases
- Tax Preparation Software: Automate tax return population by extracting W-2 and 1099 data directly into tax software.
- Lending & Underwriting: Streamline income verification processes by extracting structured financial figures from applicant tax documents.
- Accounting & Audit Firms: Perform bulk-ingestion of client tax records during high-volume tax seasons.
- Payroll Compliance Audits: Cross-reference extracted W-2 box totals against payroll records to detect discrepancies.
- Integrate with Openclaw Skills: Combine with other Openclaw Skills for advanced financial automation and document routing.
How DeepRead Tax Forms Works
- Document Upload: The agent uploads a PDF or scanned image of a tax form (e.g., W-2, 1099) via a secure POST request to the DeepRead API.
- Schema Application: DeepRead processes the document using a targeted extraction schema to align fields (like employer_ein and wages_box1) to specific boxes.
- Extraction & Confidence Flagging: The API generates typed JSON, mapping values and flagging low-confidence extractions with a needs_review boolean.
- Downstream Execution: The agent retrieves the completed payload, ready to be passed to database storage, verification engines, or other Openclaw Skills.
DeepRead Tax Forms Setup
To install and configure the DeepRead Tax Forms skill, follow these steps:
- Get your API key by visiting the dashboard:
open "https://www.deepread.tech/dashboard/?utm_source=clawhub"
- Export your API key as an environment variable:
export DEEPREAD_API_KEY="sk_live_your_key_here"
Alternatively, if you do not have an API key yet, you can authorize and fetch one via OAuth device flow using your agent:
clawhub install uday390/deepread-agent-setup
DeepRead Tax Forms Data Schema & Taxonomy
The extraction schema defines the strict structure of the resulting JSON. Below is an example schema for parsing W-2 forms:
| Property | Type | Description |
|---|---|---|
form_type |
string | Tax form type (e.g., W-2, 1099-NEC, 1040) |
tax_year |
number | Tax year on the form |
employer_name |
string | Employer or payer name |
employer_ein |
string | Employer Identification Number |
employee_name |
string | Employee or recipient name |
employee_ssn_last4 |
string | null | Last 4 digits of SSN only |
wages_box1 |
number | Box 1 — wages, tips, other compensation |
federal_tax_withheld_box2 |
number | Box 2 — federal income tax withheld |
social_security_wages_box3 |
number | null | Box 3 — Social Security wages |
medicare_wages_box5 |
number | null | Box 5 — Medicare wages |
state_wages_box16 |
number | null | Box 16 — State wages |
For 1099 documents, exchange employer/employee fields for payer/recipient keys (e.g., payer_name, nonemployee_compensation_box1).
DeepRead Tax Forms Advanced Features
- PII Redaction: Fully redact sensitive identifiers (like SSNs, EINs, and addresses) before storing or sharing files using the companion skill:
clawhub install uday390/deepread-pii. - Custom Optimization Blueprints: Achieve a 20-30% accuracy lift for recurring document types by configuring blueprints at the DeepRead Dashboard Optimizer.
- Human-in-the-Loop Validation: Filter output payloads for
needs_review: trueto flag questionable extractions for human verification automatically. - Multi-Agent Interoperability: Orchestrate complex document pipelines by chaining this tool with other Openclaw Skills like
deepread-pay-stubsordeepread-ocr.
SKILL.md
Loading
Related Openclaw Skills
Alibaba Cloud DNS Resolution Diagnostics
An automated diagnostic utility designed to troubleshoot DNS resolution anomalies, domain unreachable issues, and routing misconfigurations across Alibaba Cloud and third-party DNS platforms.
sdw-kb: Knowledge Graph Generation & Community Detection
Turn any folder of files, code, docs, images, and videos into a navigable, clustered knowledge graph with a comprehensive audit trail.
Test-Deployment Template Starter Pack
A battle-tested Python starter template to build and deploy production-ready vertical AI agents for Telegram in 10-15 minutes.
Gaokao Volunteer Assistant
An AI-powered gaokao volunteer filling helper using ranking-based matching and risk classification to automate Chinese college selection.
belief-assessor
An LLM-driven epistemic reasoning engine that evaluates claims against retrieved evidence, outputting a structured belief state with calibrated confidence levels.
MetricHub Multi-Tenant Metrics Platform
MetricHub is a multi-tenant SaaS metrics platform featuring query building, ECharts visualization, and dashboard orchestration optimized for Openclaw Skills.
Featured*
