More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Table to Excel Studio

    1

    Turn PDF, screenshot and CSV tables into editable Excel, with original files preserved, source-linked review and deliberate number formats.

    Free

    2 usesSecurity scanned
    Table to Excel Studio

    Table to Excel Studio

    Example session with this skill installed

    Fictional supplier-catalog example: convert the supplied ruled native PDF to editable Excel. Keep SKU values 00104 through 00108 as text, convert Pack to integer and Unit price to decimal, preserve the zero price, retain the original PDF and exact raw extraction, then inspect the saved workbook and deliver the evidence pack. Flag any uncertain reading instead of guessing.

    • Read your context and instructions
    • Compiled the table to excel
    • Generated the data export

    Creator-local result: tables.xlsx contains five native editable data rows, a raw table, Sources and Changes. IDs remain 00104, 00105, 00106, 00107 and 00108; quantities remain 12, 6, 24, 8 and 1; prices remain 12.50, 18.00, 7.25, 21.90 and 0.00. The pack includes per-table CSV, byte-identical original PDF with hash, extraction/dataset JSON and saved-value verification. Review/Checks appear only when applicable. This is a real saved fictional example, not a promise of verified Agensi buyer execution.

    tables.xlsx

    CSV · data export

    Generated

    Example file from a real run - the skill writes it into your workspace.

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

    About this skill

    Turn a supplier PDF, screenshot table, scanned list or CSV export into an editable workbook you can actually use. Table to Excel Studio keeps original files, makes uncertain readings visible and gives corrections a source trail.

    Complete free English edition. No seller subscription or seller-hosted backend is required. Your agent needs the listed execution, spreadsheet and file-delivery capabilities.

    From source table to useful Excel

    • Extract ruled native PDF tables, readable English screenshots/scans and UTF-8 CSV/TSV with the optional local helpers. For borderless, rotated, merged or complex layouts, use agent-assisted reconstruction against the original evidence.
    • Preserve source bytes with SHA-256, exact raw extraction, page references and cell identifiers. Mixed batches retain each source's outcome, including failures.
    • Review OCR/transcribed cells against the original. The optional local workbench provides zoom, cell highlights, cropped evidence, next-item review, reasons, undo and a separately saved review record.
    • Choose headers and number/date conventions explicitly. Keep leading-zero IDs, long references and scientific-looking codes as text. Convert quantities, amounts and dates only when their meaning is clear; unsafe precision stays exact text.
    • Inspect an export preflight with the same preparation rules as the exporter. Type errors, unreviewed cells, failed sources and failed independent totals remain visible. A clean preflight does not certify OCR accuracy.

    Workbooks made for continued use

    Receive native editable Excel cells, readable column widths, number formats, filters, frozen panes and print setup. Working data cells stay unmerged. Six restrained palettes suit finance, operations, research and everyday administration.

    The output pack can include tables.xlsx, per-table CSV, byte-identical originals, extraction/dataset JSON, cell provenance, review decisions and saved-file verification. Workbooks retain raw extracted tables and source references; Review, Changes and Checks sheets appear when relevant. Checks formulas recalculate against edited data. Review/history are export-time snapshots and do not update automatically after direct Excel edits.

    Broad practical coverage

    The catalog contains 56 fictional specimens across 12 table families: products/catalogs, inventory, purchasing/quotes, schedules, education, expenses/statements, orders, registrations, surveys, research, logistics and travel/personal administration.

    Eight additional working examples address continuation pages, an independent invoice total, European currency/grouping, small measurements, notes before headers, Unicode/multiline fields, long references/literal formula-like text and headerless input. The compact core includes 56 portable CSV examples. Full native PDF, screenshot and scanned fixtures are creator-development assets; a multi-table example's portable CSV contains only its first table and omits independent PDF-footer controls.

    What has actually been checked

    Creator-local acceptance covered 56 saved workbooks, 5,223 stored-value/raw-cell/original-hash checks, 242 saved-sheet renders and 20 behavioral regression tests. Representative LibreOffice headless open/save and formula recalculation checks passed. Real browser checks covered uploads, source review, downloads, header/locale choices, failed-file handling and a 390 px layout. These are local test results, not a universal OCR accuracy benchmark or buyer-session certification. Gallery workbook images are renders of real saved XLSX files, not Microsoft Excel screenshots; all examples are fictional.

    Runtime and honest limits

    The helper needs Python 3.11+ with the disclosed libraries, Node.js with a supported host-provided @oai/artifact-tool capability and JSZip; local OCR additionally needs Tesseract with English data. These dependencies are not bundled and are not automatically installed. If the supplied exporter is unavailable, the agent must use permitted native spreadsheet tools that meet the same evidence and verification contract, or explain the unavailable capability.

    The Skill runs in the buyer's agent. Access is through Agensi's agent connection under platform terms and service availability; this listing does not promise a downloadable/offline copy of protected Skill source. Protected scripts/resources must follow platform loading and in-session execution rules. User-generated workbooks and evidence are separate deliverables, retrieved through permitted file channels.

    Agensi buyer-side auxiliary-file loading, execution and generated-output retrieval remain unverified and require platform validation. Microsoft Excel desktop editing/save and native UI editing also remain unverified. The optional creator-local workbench is not an Agensi-hosted service. Unreadable or ambiguous content needs review; there is no claim of lossless OCR, automatic support for every layout or guaranteed data accuracy. Spreadsheet-safe CSV may prefix formula-like text; exact values and null/empty distinctions remain in JSON.

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

    Agensi buyer-side auxiliary-file loading, interpreter/module execution and generated-output retrieval remain unverified; platform validation is requested. Microsoft Excel desktop edit/save and native UI editing are unverified. Automatic helper targets readable ruled tables and English OCR, not arbitrary layout recovery. All OCR/transcribed cells require source review. No lossless OCR or universal accuracy guarantee. Review/history are export-time snapshots; direct Excel edits do not update them. CSV cannot distinguish null from empty and may prefix formula-like text; exact values remain in JSON. Core examples are portable CSV; development PDF/image fixtures are not bundled.

    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

    2 uses

    Times used to date

    Free to use

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 24 days ago

    • Free with an Agensi account

    Software
    Node.js@oai/artifact-toolJSZipTesseract
    Optional
    Python 3.11+pdfplumberpypdfium2Pillowopenpyxl

    Node.js runs the XLSX exporter; @oai/artifact-tool comes from your agent host and dependencies are not bundled. Tesseract needs English language data for local OCR.

    Permissions

    File Scopes

    Read
    only
    user-selected
    table
    sources
    and
    matching
    review
    decisions
    Write
    new
    workbooks
    CSV
    copied
    originals
    hashes
    private
    evidence
    to
    a
    chosen
    output
    location;
    never
    overwrite
    input
    files

    Optional helper binds only to loopback and validates Host/Origin. Supplied scripts do not upload sources, fetch remote URLs or install dependencies automatically. Agent image-understanding tools have their own data handling; disclose their use and follow user preferences. Originals/evidence may contain private information.

    Listed24 days ago

    What's inside

    Frequently Asked Questions

    Connect your AI agent to Agensi through MCP, then ask it to use Table to Excel Studio on your task. Your agent loads the latest approved version each time. Paid skills require a purchase or active subscription. Skill files are not available for download; your licence does not allow you to save, copy or export any part of a skill.

    The optional helper handles ruled native PDF tables, readable English ruled screenshots/scans and UTF-8 CSV/TSV. Borderless, rotated, merged, distorted or complex sources need agent-assisted reconstruction and source review. The helper limits are 32 files / 64 MiB per batch, 60 pages per PDF and 40 megapixels per image; larger inputs need an explicitly adapted workflow.

    Yes: exported XLSX contains native cells, formats, filters and unmerged working data. Images are not substituted for editable table data. Checks formulas recalculate on edited data; review/history are snapshots and do not synchronize after direct edits. Microsoft Excel desktop edit/save remains unverified.

    Exact extraction is retained separately from review decisions. OCR/transcribed cells are flagged for source review; the local workbench offers zoom, highlights, evidence crops, reasons and undo. Check counts/totals against independent controls where available. Confidence and a clean preflight do not certify accuracy; there is no 100% OCR claim.

    IDs stay text unless explicitly converted. Unsafe numeric precision stays exact text and is flagged. Original bytes are copied with hashes, raw extraction remains unchanged and corrections have source-cell IDs. CSV has transport/blank-value limits; unescaped values and null/empty distinctions remain in JSON.

    There are 56 fictional specimens across 12 families, including eight advanced examples for continuation pages, independent totals, European formats, small measurements, pre-header notes, multiline Unicode, long IDs and headerless input. The core carries portable CSV examples; full creator-development PDF/image fixtures are not included. A multi-table example's CSV contains its first table only.

    Creator-local testing passed 56 saved workbooks, 5,223 value/raw-cell/hash checks, 242 sheet renders, 20 behavioral tests, representative LibreOffice headless open/save and recalculation, plus actual browser uploads/review/downloads. These are local evidence, not a real-world OCR benchmark. Agensi buyer-side auxiliary-file loading, interpreter/module execution and generated-output retrieval remain unverified; platform validation is requested. Microsoft Excel desktop edit/save and native UI editing are unverified. Automatic helper targets readable ruled tables and English OCR, not arbitrary layout recovery. All OCR/transcribed cells require source review. No lossless OCR or universal accuracy guarantee. Review/history are export-time snapshots; direct Excel edits do not update them. CSV cannot distinguish null from empty and may prefix formula-like text; exact values remain in JSON. Core examples are portable CSV; development PDF/image fixtures are not bundled.

    New in Productivity

    Free