Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Blindspot - Simulates real users through AI agents

    1

    A multi-stage usability audit skill that uses isolated personas and independent verification to find UX blind spots.

    Free

    2 usesSecurity scanned
    Blindspot - Simulates real users through AI agents

    Blindspot - Simulates real users through AI agents

    Example session with this skill installed

    Run a quick usability audit on https://example-bakery.com. Focus on whether a first-time visitor can easily find the gluten-free options and the store hours.

    • Read your context and instructions
    • Compiled the blindspot - simulates

    I have completed a quick mode audit of example-bakery.com. I ran 3 personas through 2 targeted tasks and performed a heuristic sweep of the homepage and product listings.

    Usability Audit Report: Example Bakery

    IDSeverityCategoryFindingEvidence
    F-01HighNavigationGluten-free filter is hidden behind a generic "Refine" button not visible on mobile.S1-E04
    F-02MedTrustNo physical address or phone number in footer; trust markers missing.L1-E02
    F-03LowContentStore hours are only listed in a low-contrast image overlay on the 'About' page.S2-E09

    Scorecard Summary

    • Overall Score: 6.4/10
    • Navigation: 4.0/10 (Critical blockers found)
    • Trust & Content: 5.5/10
    • Mobile Friendliness: 7.0/10

    Next steps

    1. Move the Gluten-Free filter to the top-level navigation or breadcrumbs to reduce task friction.
    2. Add a text-based footer with the store's physical address and phone number to improve Trust scores.
    3. Replace the 'Hours' image with machine-readable text in the footer for accessibility and clarity.

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

    About this skill

    Developers often struggle to see their own work through a user's eyes. You know exactly where every button is, so you cannot objectively test if a first-time visitor can find it. This skill eliminates that builder bias by simulating a rigorous, multi-stage usability audit using isolated agents who have zero prior knowledge of your site.

    What it does

    • Persona simulation runs 2-10 distinct user profiles through specific tasks to find where they get stuck or give up.
    • Lens auditing applies structured checklists for usability heuristics, content trust, forms, mobile performance, and accessibility.
    • Evidence-based reporting links every finding to a specific log entry (e.g., S2-E03) to ensure issues are reproducible.
    • Independent verification uses a fresh agent to try and disprove every candidate finding, filtering out false positives and fetch artifacts.
    • Automated scoring generates a scorecard.json with 0-10 ratings across categories based on impact, frequency, and persistence.

    How it works

    1. Reconnaissance: A blind agent maps the site and builds a coverage ledger while the developer's brief is quarantined.
    2. Hunting: Persona sessions and lens audits run in total isolation to capture raw, unbiased visitor experiences.
    3. Validation: Findings are merged by root cause and sent to an independent verifier for a "devil's advocate" review.
    4. Reporting: The orchestrator reconciles findings against the original brief and generates REPORT.md and structured JSON data.

    Frameworks & tools

    This skill utilizes Python for validation scripts and data processing. It integrates with web browsing tools, fetch utilities, and filesystem APIs to manage the audit pipeline and record session logs.

    Why this beats prompting it yourself

    Standard prompts suffer from "model helpfulness" where the AI guesses your intent or perseveres through bad UX. This skill enforces isolation and "patience budgets" to ensure the AI fails exactly where a real human would.

    Use cases

    • Auditing a new landing page before launch to identify conversion blockers.
    • Performing a competitive teardown to see why a rival's checkout flow feels smoother.
    • Re-testing a site after UI changes to measure usability improvements via a blind delta.
    • Auditing complex signup flows to find high-dropoff friction points.

    Known limitations

    Does not perform security penetration testing, SEO keyword analysis, or legal compliance reviews. Cannot bypass bot detectors, login walls, or CAPTCHAs without provided credentials.

    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.

    Cannot bypass bot detection or CAPTCHAs. Read-only by default; does not complete purchases. Accessibility checks are expert-style, not legal WCAG certifications. Requires Python for validators.

    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 5 days ago

    • Free with an Agensi account

    Software
    PythonGit

    Permissions

    Terminal / Shell
    Network Access
    Read Files
    Write Files
    Browser
    Listed5 days ago

    What's inside

    Frequently Asked Questions

    Unlike a standard prompt that tries to be helpful, this skill uses 'quarantined' agents with limited patience budgets to simulate a first-time user's genuine confusion and friction points.

    The skill provides a comprehensive REPORT.md containing reproducible findings, a scorecard.json for automated workflows, and timestamped session logs for every simulated persona.

    You will need an environment with Python support and access to web browsing and filesystem tools to allow the agents to navigate sites and record their logs.

    Yes, you can provide credentials in your initial brief to allow the persona agents to audit authenticated areas like dashboards, checkout flows, or member-only portals.

    The skill is designed to work with any high-reasoning LLM that supports tool-calling, as the logic relies on the multi-agent pipeline rather than a specific model's internal knowledge.

    The Independent Verification stage sends a fresh agent to attempt to disprove every finding, ensuring that only consistent, reproducible UX issues make it into your final report.

    New in UX & UI Design

    Free