More screenshots

    Works with the AI Agents you already use

    Claude CodeClaude CodeCursorCursorCodex CLICodex CLIMuseMuseDotsDots+21 more

    Test Handover — an acceptance report that proves it's done

    1

    Maps every agreed feature to tests that were actually run, records expected vs actual with evidence, re-verifies fixes, and ends with a sign-off page your client can accept feature by feature.

    $5

    Secure checkout via Stripe

    Money-back guarantee

    Converts to your local currency at checkout

    Security scanned
    Test Handover — an acceptance report that proves it's done

    Test Handover — an acceptance report that proves it's done

    Example session with this skill installed

    Write an acceptance report for my client. The scope is in SPEC.md; run the tests and tell me what's really done.

    • Read your context and instructions
    • Compiled the test handover

    Bean & Kettle back office: Test and acceptance report

    Build tested back office v1.0 ·

    Environment local test copy with fictional sample data ·

    Test date 2026-11-09 ·

    For the shop owner

    Fictional example. Every result below comes from a real run of the included tests and a scripted browser walk-through.

    Summary

    Features in scope6 (from the agreed scope)
    Results2 Pass · 3 open defects · 1 moved to phase 2
    Automated tests (business rules)first run 9 passed, 3 failed → after fixes 12 passed, 0 failed
    Browser walk-through (screens)6 checks: 3 passed, 3 failed
    Defects found5: 2 fixed and re-verified; 3 open, scheduled for fixing before go-live
    Moved to phase 21: order confirmation email (F6), agreed with the client

    Verdict: F1 and F2 are ready for acceptance now. F3, F4 and F5 are accepted once defects #3–#5 are fixed and re-tested, scheduled before go-live. F6 moves to phase 2, as agreed with the client.

    Feature checklist

    IDFeatureResultHow verified
    F1Order total = sum of its lines✅ Passautomated test
    F2Discount codes (percentage, never below 0, expired refused)✅ Pass after fixautomated tests; defect #2 fixed
    F3Monthly revenue report (paid + shipped, no test accounts)❌ Open: defect #5calculation passes after fixing defect #1; the Revenue screen is not connected to it
    F4Roles: staff view orders; only owner sees revenue and edits prices❌ Open: defect #4role rules pass; sign-in accepts any password
    F5Refunds (paid/shipped only, owner only)❌ Open: defect #3refund rules pass; the refund is not saved on screen reload
    F6Order confirmation email⬜ Phase 2not in this build; moved to phase 2

    Round 1: rules and calculations

    CheckExpectedActual (first run)Result
    Order of 2 × 16.00 + 1 × 48.0080.0080.00✅
    10% code on 80.0072.0072.00✅
    120% code on 50.000 (never below 0)−10.00❌ defect #2
    Expired coderefusedrefused✅
    October revenue86 orders · 10,981.00 (independent count)92 orders · 11,674.00❌ defect #1
    Refund a paid orderstatus becomes refundedrefunded✅
    Refund a cancelled orderrefusedrefused✅

    Round 2: who can do what

    RoleActionExpectedActualResult
    staffview ordersallowedallowed✅
    staffopen revenue reportrefusedrefused ("owner-only")✅
    staffedit pricesrefusedrefused✅
    staffrefund an orderrefusedrefused ("owner-only")✅

    Round 3: back-office screens (browser)

    CheckExpectedActualResult
    Staff sees only the Orders menuno Revenue menu, no Refund buttonas expected✅
    Refund button hidden on cancelled ordershiddenhidden✅
    Owner refunds order #754status refunded, confirmation messageas expected✅
    Order #754 after reloading the pagestatus stays refundedstatus back to paid❌ defect #3
    Sign in with a wrong passwordrefusedsigned in as Owner❌ defect #4
    Revenue screen follows the revenue calculationlive figuresfigures fixed in the page❌ defect #5

    Defects

    #1 Revenue report counted refunded orders · Major · ✅ fixed

    What happened: October revenue showed

    11,674.00 from 92 orders; the correct figure is

    10,981.00 from 86 orders. The difference is exactly the 6 refunded October orders (693.00).

    • After fix: 86 orders · 10,981.00, matches the independent count ✅

    #2 Discount could make a total negative · Minor · ✅ fixed

    What happened: a code above 100% turned a 50.00 order into

    −10.00. No such code exists today, so no real order was affected.

    • After fix: total is 0.00 ✅

    #3 Refund is not saved · Major · open, fix before go-live

    • Reproduce: owner → Orders → #754 → Refund order → Yes, refund → reload the page → open #754.
    • Agreed action: save the refunded status to the shop database; re-test before go-live.

    #4 Sign-in accepts any password · Blocker · open, fix before go-live

    Security impact: yes. Anyone who knows an account email can sign in. The back office must not be published before this is fixed.

    • Agreed action: server sign-in with stored, hashed passwords; re-test before go-live.

    #5 Revenue screen shows fixed figures · Major · open, fix before go-live

    • Agreed action: connect the screen to the revenue calculation; re-test before go-live.

    Not tested

    IDFeatureCurrent stateAgreed action
    F6Order confirmation emailThis build contains no code that sends email.Built and tested in phase 2.

    Sign-off

    IDFeatureAcceptedNot acceptedComment
    F1Order total☐☐
    F2Discount codes☐☐
    F3Monthly revenue report☐☐after defect #5 is fixed
    F4Roles☐☐after defect #4 is fixed
    F5Refunds☐☐after defect #3 is fixed

    Delivered as test-handover.md plus a print-ready HTML with a signature block; a developer appendix lists suite output, causes and fixes.

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

    About this skill

    "Is everything working?" is the question that holds up your final invoice. Test Handover answers it with evidence instead of a promise.

    • Starts from the agreed scope (spec, quote, ticket list) so every feature appears, including the ones nobody tested.
    • Runs and records: your automated suite, permission checks per role, main flows, edge cases. Each check gets expected · actual · result · how it was verified.
    • Defects the client can follow: what happened, what should happen, how to reproduce, severity, and whether it's a security issue.
    • Fix round: re-runs failed checks after you fix them and shows before/after.
    • Sign-off page: Accepted / Not accepted per feature, ready to sign.
    • Developer appendix: suite output, causes and fixes, kept out of the client's way.

    The one rule: it never reports a test that wasn't run. No evidence means Not tested, with the reason. That's what makes the report worth signing.

    The included example is a real run on a small fictional shop: 12 automated tests (9 passed on the first run, 12 after two fixes) plus a browser walk-through that records three open defects, one of them a sign-in that accepts any password, each with severity, reproduce steps and the agreed action.

    Every report also comes as a styled single-file HTML, print-ready for PDF. Writes in your language (English, Chinese, Japanese and more).

    Works with Claude Code, Codex CLI, Gemini CLI, Cursor and other agents that can read files and run shell commands.

    Only as complete as your tests and environment allow; features that can't be exercised are listed as Not tested rather than guessed.

    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 8 days ago

    Requires

    Python

    Permissions

    Terminal / Shell
    Browser
    Read Files
    Write Files

    Runs your test suite and, if available, a browser walk-through against your local or test environment; writes the report into your project folder. It does not change application code while testing.

    Listed8 days ago

    What's inside

    Frequently Asked Questions

    No. Only checks it ran (or result files it can read) are reported; everything else is Not tested.

    No. It records first, then you fix, then it re-verifies, so the report reflects what was delivered.

    Yes, if your agent has browser tools or Playwright; otherwise it uses automated tests and API checks, and gives you a checklist for the steps it can't do.

    No. Test accounts are named by role, IDs masked, screenshots checked.

    A software company in Taiwan that builds, hands over and rescues business systems. Contact: youllook@gmail.com

    New in Testing & Debugging