More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIDeepSeekDeepSeekMuseMuse+22 more

    Product Writing That Tells the Truth: UI Copy, Errors, and Docs

    1

    Write the words a person meets mid-task: labels, errors, empty states, command output, and docs that say what is true and what to do next, with a scored check before they ship.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Product Writing That Tells the Truth: UI Copy, Errors, and Docs

    Product Writing That Tells the Truth: UI Copy, Errors, and Docs

    Example session with this skill installed

    I run a fictional neighborhood tool-lending app called LendShelf. We are launching a feature that lets members reserve a tool for pickup. Three pieces of copy need review: the error when a tool is already reserved, the empty state for a member with no reservations, and a line in the setup guide. Review them with the skill and give me the rewrites plus the rubric score.

    • Read your context and instructions
    • Compiled the product writing that

    DEMO — LendShelf tool reservations, invented scenario and numbers for this demo only.

    Error: the draft reads "Reservation failed. Please try again later." It names nothing and offers no next step. Rewrite: "Couldn't reserve the cordless drill — another member booked it for Saturday. The next open pickup is Tuesday. Your request is saved, so pick Tuesday or join the waitlist." The failure is named, the reader's request is accounted for, and the next step is concrete. The first version of this rewrite forgot to say the request was saved and scored 1 on the next-step line; fixed before delivery.

    Empty state: the draft reads "No reservations." It teaches nothing. Rewrite: "No reservations yet — browse tools and reserve your first pickup." Button: "Browse tools."

    Guide line: the draft reads "Pickup is quick and easy." The first rewrite claimed most pickups take under 5 minutes, but no timed run backed that number — the claim check caught it and the number was cut. Final: "Pickups happen at the front desk. Bring the confirmation code from your reservation email."

    Rubric: 2 on every line. Total 12. The copy ships.

    Claim check: the booking calendar confirmed Tuesday as the next open pickup, so the date in the rewrite is verified, not guessed. The date stays owned by the calendar, which generates it.

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

    About this skill

    Product words are read while the reader is busy, and every sentence either moves their job forward or stands in its way. This skill starts by finding the reader's moment — what they were doing, what they need next, what they fear losing — then verifies every claim against the specification, the implementation, or an observed result before a single sentence is written. From there it covers the four kinds of product copy that fail most often: interface copy where errors must open a path forward, states must say what they mean, and consequential facts must survive shortening; command-line and status output where machine formats stay clean, identifiers stay copyable, and unknown, empty, and default stay three different facts; documentation where each page takes one job, the README stays disciplined, and changeable facts get a named owner or a generated value; and the review pass that reads every piece twice, once as the user mid-task and once as the skeptic. A fill-in copy template, a six-line scoring rubric that ships copy at 10 or more out of 12 with no row at zero, a worked example that rewrites an error, an empty state, and a guide line for a fictional app, plus edge cases and a failure-mode catalog round out a method you can run on any label, alert, or page.

    Find more skills like this on the Claude Skills Marketplace, the Codex Skills Marketplace and the Gemini Skills Marketplace, or browse Claude Skills for Technical Documentation.

    • It cannot verify a claim without a source: if no spec, implementation, or observed result is available, the claim is marked, not confirmed.
    • It writes words, not product decisions: when honest copy needs the behavior to change, it flags the decision and hands it back.
    • A rubric score is a check on the copy, not a substitute for watching a real reader use it.

    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 3 days ago

    Permissions

    No special permissions declared or detected

    Listed3 days ago

    What's inside

    Frequently Asked Questions

    Mark the claim as unverified and either find the source before shipping or write the sentence so the gap is visible. Never polish a guess into a fact to make the sentence sound finished.

    No. It covers the words a person meets while using a product — labels, errors, status lines, help text, and docs. Marketing copy answers a different question and needs a different method.

    The skill flags it as a product decision and hands it back. Rewriting the sentence does not fix the behavior, and hiding the behavior behind nicer wording makes the product worse.

    New in Technical Documentation