More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Document Compare Studio

    1

    Compare document versions with source-located redlines, editable Word reports, complete change ledgers and private review drafts.

    Free

    Security scanned
    Document Compare Studio

    Document Compare Studio

    Example session with this skill installed

    Compare my supplied fictional protocol versions. Baseline: "Record observations every 10 minutes. Retain all raw readings. Use protocol version 1.2." Revised: "Record observations every 15 minutes. Retain all raw readings. Document missing readings without replacing them with zero. Use protocol version 1.3." Confirm direction, preserve both originals and produce US Letter editable reports and a complete ledger. Do not assess scientific validity or merge either source.

    • Read your context and instructions
    • Compiled the document compare studio

    The supplied protocol example yields two contiguous change groups with complete before/after text and source line locators. It flags interval 10 to 15 and version text 1.2 to 1.3, retaining the missing-reading instruction verbatim. Outputs: self-contained HTML, native editable Word comparison report, selectable US Letter PDF snapshot, CSV/JSON ledgers, exact originals, hashes and a private project. Notes remain pending until applied. No protocol-validity or implementation claim is made.

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

    About this skill

    Two versions. One clear review.

    Compare a confirmed baseline and revised document, keep both originals, and turn visible text differences into a useful review record. A complete free workflow for business, technical work, research and everyday documents.

    What you can create

    • Source-located side-by-side comparisons, inline redlines and a review overview in self-contained HTML.
    • Native editable Word comparison reports and selectable A4/US Letter PDF snapshots.
    • Complete CSV/JSON change ledgers, raw whitespace-suppression records and exact original files with hashes.
    • A private editable project with applied reviews, pending review drafts and operation history.

    A practical review experience

    Navigate with previous/next, jump or J/K. Filter numeric and requirement cues, inspect exact before/after tokens, show whitespace and open neighbouring original blocks. Review notes stay as private drafts until explicitly applied; save/reopen the project and apply one or all. A mode or source change starts a fresh review identity. Undo can restore the preceding project state.

    Overview exclusions and search affect the screen, not the complete evidence: CSV/JSON and complete printing retain every change group. Checked means reviewed; it does not merge an edit or approve the underlying claim.

    48 complete original examples

    Twelve business pairs cover scopes, quotes, budgets, procurement, purchase orders, service plans and handovers. Twelve technical pairs include releases, runbooks, requirements, permissions, logging and a longer migration plan. Twelve research pairs cover methods, denominators, data-table units, eligibility, analysis plans and reviewer responses. Twelve everyday pairs cover travel, events, household plans, notices and unchanged-file controls. Twelve visual foundations and three HTML review views support different review contexts; the agent adapts them to the actual sources.

    Preserve the evidence

    Strict comparison is the default. Whitespace-relaxed mode keeps suppressed raw differences in a separate ledger and never normalises numbers, case or punctuation. Original bytes are retained, and comparison opcodes reconstruct both complete extraction streams. Numeric and wording flags request attention; they do not certify correctness, legal effect or implementation status.

    Execution and scope

    Core text/HTML/CSV/JSON requires Python 3.10+ standard library. Optional python-docx, pypdf and ReportLab support native imports/exports. The optional editor binds to loopback. No paid API or built-in document-upload service is required.

    TXT/MD/CSV/JSON/LOG/RST are compared as decoded UTF-8 text. DOCX compares body paragraphs and table text; PDF compares native extracted text. Formatting, figures, signatures, comments and other declared omitted content need separate inspection. Image-only PDFs and DOCX tracked insertions/deletions/moves are rejected by the helper. This is not native Word Track Changes, a source merge, OCR or whole-document visual equivalence testing.

    Agensi supplies the latest Skill through its agent connection, not an offline Skill download. Generated comparison works are distinct from reusable Skill files. Buyer-side multi-file loading, permitted unchanged temporary helper execution, relative assets, optional dependencies and generated-work retrieval have not been verified; platform verification is requested before approval. Microsoft Word desktop edit/save/reopen, external reviewer acceptance and physical printing also remain unverified. The reports and private pack contain full source text; confirm sharing scope before distributing them.

    Text-scoped comparison, not native Word Track Changes, OCR, source merging or full visual/formatting equivalence. DOCX non-body elements and PDF visual content need separate inspection. Image-only PDFs and DOCX tracked insertions/deletions/moves are rejected by the helper. Text input requires UTF-8. Reports and private packs contain full source text. Word desktop edit/save/reopen, physical print, external reviewer acceptance and Agensi buyer execution/retrieval remain unverified.

    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 any payment.

    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

    Free forever

    No account required to browse

    Trust & safety

    Security scan

    Security scan passed 22 days ago

    • Free with an Agensi account

    Requires

    Python 3.10+

    Permissions

    File Scopes

    Confirmed
    user-provided
    baseline
    and
    revised
    source
    files
    New
    user-approved
    output
    revision
    directories

    Local processing. Optional editor binds to 127.0.0.1. No built-in external upload service or paid model API. Dependency installation is a separate host-environment action. Follow platform persistence rules for reusable Skill files.

    Listed22 days ago

    What's inside

    Frequently Asked Questions

    Yes. The full workflow and all 48 original paired examples are included. No paid generation API or hosted backend is required by this package. Your connected agent and its permitted execution tools are still required.

    UTF-8 TXT, MD, CSV, JSON, LOG and RST compare as text lines. Optional dependencies support DOCX body/table text and native PDF text. This is not a cell-key spreadsheet comparison. Image-only PDFs are rejected; tracked DOCX revisions require a separately confirmed clean view.

    DOCX reports contain native text, paragraphs and headings, not page images. Local rendering was checked; Microsoft Word desktop editing/save/reopen remains unverified. Reports are not native Track Changes or merged source files.

    No. Declared extraction scope omits content such as figures, layout, signatures and non-body document parts. No text changes only applies within that scope. Matching original hashes support the narrower statement that original byte streams match.

    Strict mode is default. Relaxed mode retains full suppressed raw whitespace differences separately. Numbers, case, punctuation and units are not normalised. Raw numeric/requirement cues request review, not financial, scientific or legal conclusions.

    Pending notes stay as private in-session drafts. Save the project to retain them across sessions; apply one or all before they enter reports. Changing mode/source starts a fresh review identity and keeps before/after note maps in operation history. Save the earlier project to retain its complete old sources.

    Full reports/ledgers, exact originals, extraction records, project, review history and a file manifest. Reports and packs contain full source text. Screen filters/overview exclusions do not delete complete evidence or printed change groups. Confirm sharing scope before distribution.

    Agensi delivers the latest Skill through its agent connection. No offline Skill download is promised. Buyer-side multi-file/helper loading, relative assets, optional dependencies and generated-work retrieval remain unverified; platform verification is requested before approval. Generated works are distinct from reusable Skill files.

    New in Productivity