Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Website Art Director Review

    1

    - **Evidence for every issue.** No evidence, no issue.

    $19

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Website Art Director Review

    Website Art Director Review

    Example session with this skill installed

    Review this proposed sitemap before I show it to the client:

    Hartwell Legal - proposed homepage (single page)

    Goal from the brief: "get more consultation requests from small business owners".

    Sections, top to bottom:

    1. About the firm: founded in 2009, three partners, commercial law focus.
    2. Practice areas: contracts, employment disputes, business formation.
    3. Our team: partner bios with photos.
    4. Testimonials: three quotes from business owners (already collected).
    5. FAQ: fees, timelines, first meeting.
    6. Footer: address, phone, social links.

    Notes from the designer: "Calm and serious. Navy and white. No buttons in the body, we want it to feel like a brochure, not a sales page."

    • Read your context and instructions
    • Compiled the website art director

    Art-direction review complete. Score and verdict computed by score_review.py; output validated by validate_review.py.

    {
      "schema_version": "1.0",
      "language": "en",
      "reviewed": {
        "materials": [
          "sitemap"
        ],
        "business_goal": "Get more consultation requests from small business owners",
        "cta_objective": "lead_generation",
        "sections_seen": [
          "about",
          "services",
          "team",
          "testimonials",
          "faq",
          "footer"
        ]
      },
      "score": 82,
      "approval_status": "rejected",
      "dimension_scores": {
        "strategic_alignment": 50,
        "visual_identity_fit": 100,
        "content_hierarchy": 75,
        "conversion_clarity": 50,
        "component_fit": 100,
        "asset_usage": 100,
        "consistency": 100
      },
      "issues": [
        {
          "id": "ISS-001",
          "check_id": "HIE-01",
          "severity": "critical",
          "area": "hierarchy",
          "target": "hero",
          "problem": "The page has no hero: it opens directly with the firm's history.",
          "evidence": "Section 1 of the sitemap is 'About the firm (founded in 2009...)'; no opening section states what the firm does or for whom.",
          "recommendation": "Add a hero that states the offer for small business owners in one line, with the consultation CTA."
        },
        {
          "id": "ISS-002",
          "check_id": "CNV-01",
          "severity": "critical",
          "area": "conversion",
          "target": "contact",
          "problem": "The goal is lead generation, but there is no contact or consultation section.",
          "evidence": "Goal: 'get more consultation requests'. Sections 1-6 contain no form or contact block; contact data appears only in the footer.",
          "recommendation": "Add a closing consultation section with a short form (name, company, issue) before the footer."
        },
        {
          "id": "ISS-003",
          "check_id": "CNV-02",
          "severity": "critical",
          "area": "conversion",
          "target": "page",
          "problem": "No CTA exists anywhere on the page.",
          "evidence": "Designer note: 'No buttons in the body'. No section in the sitemap includes a call to action.",
          "recommendation": "Use one primary CTA ('Book a consultation') in the hero and the closing section. A single sober button keeps the brochure feel."
        }
      ],
      "strengths": [
        "Testimonials from the target audience are already collected.",
        "The FAQ addresses fees and timelines, the main objections before a first legal meeting.",
        "The team section with partner bios and photos supports trust for a professional service."
      ],
      "recommendations": [
        {
          "id": "AD-001",
          "priority": "high",
          "type": "layout",
          "fix_layer": "structure",
          "target": "hero",
          "recommendation": "Add a hero that states the offer for small business owners in one line, with the consultation CTA.",
          "reason": "The page has no hero: it opens directly with the firm's history.",
          "requires_rework": true,
          "issue_ids": [
            "ISS-001"
          ]
        },
        {
          "id": "AD-002",
          "priority": "high",
          "type": "cta",
          "fix_layer": "strategy",
          "target": "contact",
          "recommendation": "Add a closing consultation section with a short form (name, company, issue) before the footer.",
          "reason": "The goal is lead generation, but there is no contact or consultation section.",
          "requires_rework": true,
          "issue_ids": [
            "ISS-002"
          ]
        },
        {
          "id": "AD-003",
          "priority": "high",
          "type": "cta",
          "fix_layer": "strategy",
          "target": "page",
          "recommendation": "Use one primary CTA ('Book a consultation') in the hero and the closing section. A single sober button keeps the brochure feel.",
          "reason": "No CTA exists anywhere on the page.",
          "requires_rework": true,
          "issue_ids": [
            "ISS-003"
          ]
        }
      ],
      "blocking_issues": [
        "The page has no hero: it opens directly with the firm's history.",
        "The goal is lead generation, but there is no contact or consultation section.",
        "No CTA exists anywhere on the page."
      ],
      "summary": "Score 82/100 - rejected: the plan has no hero, no consultation section and no CTA, so a lead-generation site cannot convert as designed.",
      "limitations": [
        "Only a sitemap was reviewed: copy, visuals and components (CPY, VIS, CMP checks) could not be evaluated.",
        "The brand style ('calm and serious, navy and white') was not checked against any visual material."
      ]
    }
    

    Next steps: fix the critical issues first (by layer), then re-run the review.

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

    About this skill

    by FS Studio

    A senior art director's review of any landing page or site proposal, with a score you can reproduce and defend.

    Ask an AI "what do you think of this landing page?" and you get a different opinion every time. This skill replaces opinion with a fixed review system:

    • 34 checks across conversion, strategy, structure, hierarchy, visual identity, consistency, components, assets and copy. Each check has a fixed severity.
    • Evidence for every issue. No evidence, no issue.
    • A deterministic score. A script computes 7 dimension scores, a 0-100 score and the verdict: approved, needs revision or rejected. The same issues always give the same result, and the validator rejects any review whose score doesn't match the formula.
    • Prioritized recommendations. Each one says what to change and in which layer to fix it: strategy, structure, visual, copy, assets or components.
    • Honest limitations. If you only provide HTML, the review says that visual checks were not evaluated, instead of guessing.

    Works in Spanish and English. Accepts live URLs (if your agent can browse), HTML, screenshots, sitemaps, wireframes, blueprints, copy decks, briefs and brand guidelines.

    Use cases

    • Before showing a site to the client: catch placeholder text, a hero without a CTA or a lead-generation page with no contact form.
    • Before building: review the sitemap or blueprint and find structural problems while they are still cheap to fix.
    • Agencies: a consistent quality gate across designers and projects.
    • Freelancers: a professional, client-presentable review document.

    What you get

    FilePurpose
    SKILL.mdThe skill instructions (5-step workflow plus quality gates)
    reference/review-checks.mdThe catalog of 34 checks (when each applies, when it fails, the typical fix)
    reference/checks.jsonMachine-readable rules: penalties, dimension mapping, thresholds
    reference/art-direction-review.schema.jsonThe JSON contract (v1.0)
    scripts/score_review.pyDeterministic scoring: dimension scores, verdict, recommendations
    scripts/validate_review.pyValidator: schema, catalog conformity and formula consistency
    examples/3 reviews, one per verdict: rejected sitemap (en), landing needing revision (es, with Markdown report), approved booking page (en)
    tests/25 unit tests for the scoring and the validator

    How the score works

    • Each of the 7 dimensions starts at 100.
    • Each issue subtracts a fixed penalty (critical 25, high 15, medium 8, low 3) from the dimensions mapped to its area.
    • The overall score is the rounded average of the 7 dimensions.
    • Rejected if there is any critical issue or the score is below 70.
    • Needs revision if there is any high issue or the score is below 90.
    • Approved otherwise.

    A single critical problem, such as a lead-generation site with no way to contact the business, rejects the proposal even if everything else is polished. That is intentional.

    Requirements

    • An AI agent that supports Agent Skills (SKILL.md).
    • Optional: Python 3.9+ for deterministic scoring and validation. The scripts use the standard library only and make no network calls. Without Python, the agent applies the documented formula by hand.

    Implementation status

    FeatureStatus
    Catalog of 34 checks with fixed severities✅ Included
    Deterministic scoring and verdict✅ Included, as a script with tests
    Recommendations: merged, prioritized, max 10, with fix layer✅ Included
    Strengths and limitations✅ Included
    Validator (schema, catalog and formula)✅ Included, with 25 unit tests
    Markdown report for the client or team✅ Included (generated by the agent)
    Editing or fixing the site❌ Not included. The skill only reviews
    Performance, SEO, Core Web Vitals❌ Not included
    Automated accessibility audit❌ Not included. Visible problems (such as contrast) can be reported as CUSTOM issues
    Live URL fetching⚠️ Depends on your agent's browsing tools

    Compatibility and testing

    Field-tested on 2026-11-03 with the final release. Every test gives the agent a normal user request without naming the skill. The output is checked with the skill's own validator plus case-specific expectations: correct classification, no invented data, the right verdict and so on.

    EnvironmentStatusEvidence
    Claude Code (CLI 2.1.317)✅ Functionally tested6/6 cases passed. The skill was discovered and loaded automatically from .claude/skills/ in every run
    OpenAI Codex CLI (0.164.0)✅ Functionally tested6/6 cases passed. The skill was discovered automatically from .agents/skills/ in every run

    Built on the open Agent Skills standard: a plain SKILL.md with name/description frontmatter, relative paths only and no tool-specific features.

    What the 6 cases cover: 4 cases use inputs that are not in the shipped examples, so the agent cannot copy an example. The other 2 reproduce the shipped examples and only show that the examples can be reproduced. The cases include realistic inputs and edge cases: incomplete input, ambiguous input, unusual wording and mixed languages.

    Validator and unit tests: these run anywhere with Python 3.9+ (python -m unittest discover -s tests).

    AI agents are not deterministic. Two runs on the same input can phrase things differently or find slightly different sets of minor issues. The contract, the closed vocabularies and the validator keep the structure and the critical decisions consistent.

    The score is deterministic for a given set of issues: the same issues always produce the same score and verdict. Different agents reviewing the same material agreed on the verdict and the critical issues in every test, but sometimes reported a different number of minor issues, so the final score can differ by a few points.

    Pricing

    US$19: a single license for your own and your clients' projects. It will also be part of the Web Proposal Pack for US$39.

    License

    FS Studio Commercial Skill License (single user). See LICENSE.txt.

    • ✅ Use it in personal and commercial projects, including paid client work.
    • ✅ The outputs you generate (blueprints, reviews, change requests) are yours.
    • ❌ No redistribution, sharing or resale of the skill or its files, alone or bundled in another product.

    The license supplements the Agensi Terms of Service, which prevail in case of conflict.

    © 2026 Facundo Sicart · Published by FS Studio

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for UX & UI Design.

    Reviews and recommends; does not fix the site. Does not measure performance, SEO or Core Web Vitals, and is not an automated accessibility audit. Reading a live URL depends on your agent's tools. The score is deterministic for a given set of issues; two agents may find different minor issues, so scores can vary by a few points (verdict and critical issues matched in testing).

    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 14 days ago

    Nothing to install. It runs in your AI agent.

    Optional
    Python 3.9+

    Python only runs the bundled validator, with the standard library.

    Permissions

    Terminal / Shell
    Read Files
    Write Files
    Listed14 days ago

    What's inside

    Frequently Asked Questions

    A landing page, a full site, a sitemap or a site blueprint, from HTML, a description, a URL (if your agent can read it) or a plan.

    34 fixed checks with fixed severities. A deterministic script computes the 0-100 score and the verdict (approved / needs revision / rejected), and a validator checks the format, the catalog and the formula.

    Yes. Every issue must cite evidence from the reviewed material, and the review states honestly what it could not evaluate.

    Spanish and English.

    It is field-tested in Claude Code and OpenAI Codex CLI: every test case passed in both, using normal requests that never name the skill, and the skill loaded automatically. It is a plain SKILL.md skill (open Agent Skills standard) with no tool-specific features.

    No. Python 3.9+ is optional and only used to run the bundled validator (standard library only, no network access, no dependencies). Without Python, the agent applies the same quality gates manually.

    Yes. The license covers personal and commercial use, including paid client work, and the outputs you generate are yours. Redistributing or reselling the skill files is not allowed.

    New in UX & UI Design