Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Design Md Evidence Extractor Pro

    2

    Turn a user-approved website into a provenance-backed DESIGN.md pack that separates measured tokens, inferred roles, conflicts, and unresolved gaps.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Design Md Evidence Extractor Pro

    Design Md Evidence Extractor Pro

    Example session with this skill installed

    Extract an evidence-backed DESIGN.md from https://example.com using the homepage, pricing page, and one product page. Do not copy text or images. Save the result under ./output and tell me whether coverage is READY or PARTIAL.

    • Read your context and instructions
    • Compiled the design md evidence

    Creates ./output/DESIGN.md, design-tokens.json, extraction-evidence.json, coverage-report.md, and rebuild-brief.md. The coverage report states READY, PARTIAL, or BLOCKED with a score, missing groups, conflicts, warnings, and the next safe capture action. Every proposed token keeps page-level provenance in extraction-evidence.json.

    design-tokens.json, extraction-evidence.json, coverage-report.md

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

    About this skill

    Design.md Evidence Extractor is built for frontend developers, agencies, design-system teams, and AI coding workflows that need more than a guessed palette. Its Evidence-Lock Extraction Pipeline captures rendered computed styles from representative pages, preserves page-level provenance, detects conflicting variables and component variants, scores coverage, and compiles a five-file design intelligence pack. The output includes DESIGN.md, normalized design tokens, an evidence ledger, a coverage report, and a rebuild brief. Thin pages, blocked stylesheets, bot challenges, missing states, and inconsistent themes are not hidden: the skill returns READY, PARTIAL, or BLOCKED with a specific recovery path. It intentionally excludes page copy, cookies, form values, images, logos, proprietary scripts, and other source assets. Use it to document a legacy UI, create an authorized migration baseline, prepare agent-ready visual context, compare page families, or reduce unsupported design guesses before implementation.

    The skill requires a runtime that can evaluate JavaScript in the rendered page; text-only browsing is insufficient.
    Bot challenges, login walls, inaccessible pages, and heavily isolated embedded content can produce PARTIAL or BLOCKED results.
    Hover, focus, mobile, dark-mode, animation, and authenticated states are included only when explicitly captured.
    Semantic role assignment is conservative but still partly inferential; the evidence ledger records the basis and confidence.
    The upstream DESIGN.md format is alpha, so future specification changes may require an update.
    This skill documents authorized visual facts; it does not grant rights to copy trademarks, content, assets, code, or a complete website.

    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

    Money-back guarantee

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

    Trust & safety

    Security scan

    Security scan passed 3 months ago

    Software
    Python@google/design.mdNode.js

    Permissions

    Terminal / Shell
    Write Files
    Network Access
    Read Files
    Browser

    File Scopes

    ./input/**
    ./work/evidence/**
    ./output/**

    Browser and network access capture rendered styles only from user-approved public domains. Read Files loads the bundled collector and evidence files. Write Files stores temporary JSON evidence and the five output artifacts. Terminal / Shell runs the bundled standard-library Python compiler. The skill does not require environment variables, credentials, cookies, broad filesystem access, or automatic package installation.

    Listed3 months ago

    What's inside

    Frequently Asked Questions

    It solves the 'design guess' problem by auditing rendered computed styles to extract evidence-backed tokens and roles, ensuring your UI documentation is based on actual browser data rather than subjective interpretation.

    You receive the core extraction skill and a standard five-file output template: a DESIGN.md summary, normalized tokens, an evidence ledger, a coverage report, and a structured rebuild brief for your development workflow.

    The skill is designed to run within any LLM-based agent or automated workflow that supports web-browsing capabilities and markdown file generation.

    To protect privacy, the skill intentionally ignores page copy, images, logos, cookies, and form values, focusing exclusively on design tokens and structural patterns.

    Updates are released periodically to improve token normalization and edge-case detection; your marketplace purchase grants you access to all future version increments of the skill.

    Each report provides a status of READY, PARTIAL, or BLOCKED, alerting you to specific issues like bot-protection challenges or missing UI states that may require manual intervention.

    New in Frontend & Web Apps