More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    UI State Coverage Checker

    1

    Review UI interface state coverage from screenshots and specs to identify missing error, loading, and empty states.

    Free

    4 uses5.0 (1 review)
    UI State Coverage Checker

    UI State Coverage Checker

    Example session with this skill installed

    Review request

    Use the UI State Coverage Checker to review this feature before design handoff. Produce a practical state matrix and the most important follow-ups. Do not change the design.

    Fictional product: Relayboard, an authenticated team workspace. Scope: team invite form and member search. Review design documentation only.

    Source A — Invite form specification v0.3

    • A1. The email field is required. Invalid email shows "Enter a valid email address" beside the field.
    • A2. Selecting Send invite starts an asynchronous request. While pending, the button reads "Sending..." and is disabled.
    • A3. A successful request displays "Invitation sent" and adds the invite to the pending-invites list.
    • A4. The failure treatment is not specified in this document.

    Source B — Member search specification v0.3

    • B1. Search requests member records asynchronously after the user submits a query.
    • B2. When a query matches nothing, show "No matching members" with a Clear search action.
    • B3. Only workspace admins can send invitations. Non-admins see "Ask a workspace admin to invite a teammate."
    • B4. Search-loading and request-failure states are not described in these supplied pages.

    Source C — Product-owner note for this handoff

    • C1. The invite-request failure state is required but is not designed yet in v0.3.
    • C2. Session-expiry behavior during an invite request has not been decided.
    • C3. Do not assume that sending the invitation again is safe after a timeout; engineering must confirm the request outcome and deduplication behavior.

    No screenshots, prototype, implementation, network trace or runtime tests are supplied.

    • Read your context and instructions
    • Compiled the ui state coverage

    Abridged from a recorded Codex review of a fictional design specification. No live product was inspected. EVIDENCED means documented intent in this example, not verified implementation.

    IDSurface / stateCoveragePriority
    ST-03Invite / request pending and repeat-submit gatingEVIDENCED—
    ST-05Invite / request known to have failedCONFIRMED GAPP1
    ST-06Invite / timeout or unknown outcome and safe recoveryNOT EVIDENCEDP1
    ST-09Member search / request loadingNOT EVIDENCEDP2
    ST-10Member search / zero matching results and clear actionEVIDENCED—
    ST-11Member search / request fails and recoveryNOT EVIDENCEDP2

    Highest-priority decision

    The product-owner note explicitly confirms that the required invite-request failure state is not designed in v0.3. This is a confirmed design gap only. Search loading and failure are not described in the supplied pages; their absence from these pages is not proof of absence from the product.

    Before offering a repeat invitation after timeout, confirm request-outcome and deduplication behavior. A request with an unknown outcome must not be treated as a definite failure.

    Proposed acceptance checks

    • Given a response confirming failure, show failure feedback and an approved next action while preserving the entered email.
    • Given a timeout with an unknown outcome, preserve relevant input and show the approved uncertainty treatment without inviting an unverified repeat action.
    • Distinguish a failed search request from a successful query with no matches.

    These are proposed design acceptance checks, not tests run or approved new requirements. The complete report includes the remaining states, evidence references, open questions and exclusions.

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

    About this skill

    Review UI states before design handoff without treating incomplete evidence as a confirmed product defect.

    UI State Coverage Checker reviews supplied screenshots, screen descriptions and specifications. It separates documented or visible behavior from partial evidence, unresolved questions, confirmed gaps and conflicting requirements.

    What you get

    • A screen-by-screen state coverage matrix with source references.
    • Relevant checks for loading, empty, error, success, disabled, permission and recovery states.
    • Up to five prioritized follow-ups with proposed acceptance checks.
    • A clear record of unknowns, excluded states and review limitations.

    Who it is for

    Product designers, frontend developers and small SaaS teams preparing a feature for implementation or reviewing a design handoff.

    How it works

    1. Supply a screenshot, a screen description or the relevant specification. Include the user task and any known requirements.
    2. The skill identifies which states apply and classifies the evidence for each one.
    3. Review the matrix, resolve the most important questions and use the proposed checks in your handoff.

    Evidence comes first

    A missing spinner in one screenshot does not prove loading feedback is absent from the product. A specification describes intended behavior; it does not prove the implementation works. The report keeps those distinctions explicit.

    Requirements

    Use an AI agent that can load SKILL.md instructions and read the material you supply. Screenshot inputs require image understanding. No Figma connection, API key, runtime installation or additional paid service is required by the skill.

    Limits

    This is a review of supplied state evidence. It does not inspect a live product, run application tests, edit designs, certify accessibility or guarantee that every edge case has been found. Evaluation to date is in Codex; other agent hosts have not been verified.

    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 supplied evidence only; it does not inspect or test a live application.
    • Static screenshots cannot establish interaction success, timing, focus behavior or implementation completeness.
    • A design gap is kept separate from a confirmed implementation defect.
    • No accessibility certification, visual-quality audit or exhaustive edge-case guarantee.
    • Evaluation to date is in Codex only; other agent hosts and automatic trigger behavior are 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 buying a skill.

    Reviews

    5.0
    1 review
    5
    1
    4
    0
    3
    0
    2
    0
    1
    0

    This skill has been used 4 times.

    Trust & safety

    Security scan

    Security scan passed 24 days ago

    • Free with an Agensi account

    Permissions

    Read Files
    Write Files

    File Scopes

    Only
    the
    exact
    screenshots
    and
    design
    documents
    selected
    or
    supplied
    by
    buyer

    Read Files is needed only when input is supplied as local/attached files; pasted text needs no file access. The skill returns the report in conversation. Optional file output requires a separate user request and a selected destination. No shell, browser, network, credentials or service access is required by the review procedure.

    Listed24 days ago

    What's inside

    Frequently Asked Questions

    Supply screenshots, a bounded screen description or the relevant design specification. Include the user task and known requirements. Missing context is labeled or requested rather than invented.

    The skill reviews material you provide. It does not connect to Figma, browse a live product or require an API key. Your agent must be able to read your chosen input format.

    Only direct supporting evidence can establish a confirmed gap, and its scope is stated. A state not shown in a screenshot or specification is reported as not evidenced, not automatically missing from the product.

    A Markdown report in your agent conversation with a source inventory, state coverage matrix, prioritized follow-ups, proposed acceptance checks and limitations. It does not write files unless you ask for a file.

    The agent provides a structured coverage matrix, an applicability assessment for each state, and a prioritized list of action items (P1-P3) to help you fix the most critical gaps first.

    Task executions were evaluated in independent Codex contexts. The package uses plain SKILL.md instructions, but Claude Code, Cursor and other hosts have not been verified. Screenshot review requires an image-capable agent.

    No. It reviews the supplied state evidence. It does not run your app, test interactions, certify accessibility, modify designs or guarantee complete edge-case coverage.

    New in UX & UI Design

    Free