Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    document table and entity extraction

    1

    Manually mapping business documents to internal schemas is slow and prone to transcription errors.

    $2.99

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    document table and entity extraction

    document table and entity extraction

    Example session with this skill installed

    Extract the vendor name, invoice date, and line item table from this text. Use the 'finance' access scope. Invoice #INV-2024-001 from Acme Corp, dated 2024-05-15. Items: 2x Widgets at 50.00 each.

    • Read your context and instructions
    • Compiled the document table

    I have processed the document text according to the provided schema. The extraction was completed with a reconciliation check on the line items.

    Document Extraction — Internal Finance, 2024-05-22

    1. Scope, sources and assumptions

    • Documents: E1 (Invoice, Approved)
    • Schema: finance-invoice-v1
    • Assumptions: Date format MDY based on document context.
    • Access Scope: finance-access-level-1

    2. Extracted entities

    DocumentEntityFieldValueUnitStatusConfidencePointer
    E1VendornameAcme Corpn/avalid1.00E1 p1 l1
    E1Headerdate2024-05-15datevalid1.00E1 p1 l2

    3. Extracted tables

    Table: Line Items

    IDDescriptionQtyUnit PriceTotalStatusConfidence
    1Widgets250.00100.00valid0.98
    Total100.00Reconciled

    4. Exception queue

    The exception queue is empty.

    5. Confidence and reconciliation

    • Field Acceptance Rate: 100%
    • Reconciliation: Computed sum (100.00) matches stated total (100.00). Difference: 0.00.

    6. Warnings and limits

    • Code W-01: Native text layer used; no OCR artifacts detected.

    7. Items for human review and release

    • Analytical estimate — validation required: The interpretation of the date format (MDY) should be verified.
    • Draft Release: results.release generated. Note: requires_human_approval: true.

    Next steps

    1. Verify the date normalization matches your accounting period.
    2. Approve the draft release to trigger the ERP export.

    Connects securely to your tools. The creator never sees your data.

    About this skill

    Manually mapping business documents to internal schemas is slow and prone to transcription errors. This skill automates the extraction of structured entities and complex tables from document text while maintaining strict provenance and data integrity.

    What it does

    • Entity mapping transforms raw document text into typed JSON or CSV fields based on your specific schema.
    • Table extraction identifies and parses delimited tables, including row-level checks and total reconciliation.
    • Data normalization standardizes dates, currencies, percentages, and numbers according to regional conventions.
    • Ontology resolution matches extracted terms against supplied reference lists to ensure nomenclature consistency.
    • Exception handling generates a structured queue for ambiguous or low-confidence values requiring human review.
    • Access enforcement withholds sensitive values and excludes documents outside defined classification scopes.

    How it works

    1. Register documents with IDs, classifications, and text content in a request.json file.
    2. Define the target schema, including field types, validation constraints, and ontology references.
    3. Execute the provided run.py script to perform deterministic parsing, scoring, and reconciliation.
    4. Review the generated envelope to identify completed extractions, warnings, or items in the exception queue.

    Frameworks & tools

    This skill relies on a self-contained Python 3 script for processing. It uses standard JSON schemas for input/output validation and is compatible with any text-based document sources, including native PDF text or OCR output.

    Why this beats prompting it yourself

    Generic prompting often leads to "hallucinated" data or inconsistent formatting. This skill uses a deterministic script to ensure every value has a provenance pointer to a specific page and line, while performing mathematical reconciliation that LLMs typically struggle with.

    Use cases

    • Extracting line items and header data from invoices for ERP entry.
    • Mapping regulatory report data to internal compliance schemas.
    • Converting legacy forms into structured, searchable data records.
    • Reconciling multi-page tables where totals must be validated across line items.

    Known limitations

    Does not support direct image or PDF file reading; requires extracted text as input. Limited to structured extraction rather than general summarization or sentiment analysis.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse All Claude Skills.

    Requires pre-extracted text (not raw images/PDFs). No support for text analytics, sentiment, or legal/fraud verdicts. Human approval required for all releases.

    How to use

    1. 1

      Connect your agent to the Agensi MCP

      Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup

    2. 2

      Sign in to Agensi

      Your agent shows a sign-in link. Sign in and click Allow.

    3. 3

      Ask for what you need

      Your agent loads this skill and uses it on your task. It always asks before buying a skill.

    Reviews

    No reviews yet

    Be one of the first to try it. Every listed skill goes through the trust checks below.

    Security scanned

    Checked by our 8-point scan before listing

    Fresh listing

    Recently published to Agensi

    Money-back guarantee

    Doesn't do what its page promises? You get your money back.

    Trust & safety

    Security scan

    Security scan passed 2 days ago

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Environment Variables
    Listed2 days ago

    What's inside

    Frequently Asked Questions

    This skill requires raw text input. While it cannot 'read' a PDF or image directly, you can feed it text generated from any OCR engine or PDF-to-text tool to perform the extraction and reconciliation.

    Unlike standard LLM prompting, this skill uses deterministic logic to cross-check table row totals against headers. It flags any mathematical discrepancies in an exception queue for human review.

    The skill is delivered as a self-contained Python 3 script. You simply need to provide a `request.json` containing your document text and a schema defining the fields you want to extract.

    Yes, the skill includes an ontology resolution feature that compares extracted terms against your specific reference lists to ensure the output matches your internal nomenclature.

    You receive the core Python processing script, documentation on input/output schemas, and example templates for invoices and regulatory reports to help you get started immediately.

    The skill produces a detailed 'envelope' output which includes confidence scores for every field and specific provenance pointers (page/line) for every piece of data extracted.