Skills · Code

CodeSkillDugout-created

/test-scenarios

Writes short test scenarios a tester can run from the text alone, each with setup, action, expected result and status.

Created by Piyush Baheti · Source: piyush-tkd/skillsdugout/tree/main/plugins/test-scenarios

Who it is for

Developers and analysts handing a change to QA, and testers writing up what to check.

When to use it

  • A fix or feature needs checking before release.
  • You want to tell QA what has and has not been run.
  • Test notes have grown into analysis nobody needs.

Example

You ask

Write test scenarios for the new refund rule: refunds over $100 need manager approval; $100 or less go straight to the payment provider. The under-$100 case passed in staging. Over-$100 hasn't been run — staging has no manager accounts.

Without the skill

Test Scenarios: Refund Approval Rule

Rule: Refunds over $100.00 need manager approval. Refunds of $100.00 or less go straight to the payment provider.

Status

Area Status
Under-$100 refund Passed in staging
Exactly $100.00 refund Not run. "Under $100" does not cover the "$100 or less" boundary.
Over-$100 refund (approval path) Not run. Staging has no manager accounts.
Approval rejection, permissions, edge cases Not run

Action required: Create at least two manager accounts and one non-manager account in staging. Without them, every scenario in Group B is blocked, and the rule is unverified for refunds over $100.

Group A: Direct Refund (no approval)

ID Scenario Expected result Status
A1 Refund of $50.00 Sent to the payment provider. No approval record is created. Passed
A2 Refund of $100.00 (boundary) Sent to the payment provider. No approval is requested. To run
A3 Refund of $0.01 (minimum) Sent to the payment provider. To run
A4 Refund of $0.00 or a negative amount Rejected with a validation error. Nothing is sent. To run
A5 Partial refund of $100 or less on a $500 order Sent directly. The rule applies to the refund amount, not the order total. To run

Group B: Manager Approval (blocked until accounts exist)

ID Scenario Expected result Status
B1 Refund of $100.01 (boundary) Held as "Pending approval". Nothing is sent to the provider. Blocked
B2 Refund of $500.00, manager approves Status moves to approved, then the refund is sent to the provider once. Blocked
B3 Refund of $500.00, manager rejects Status is rejected. Nothing is sent. The requester is notified. Blocked
B4 Refund of $500.00, no action taken Stays pending. It is not sent automatically. If a timeout or escalation is specified, test it. Blocked
B5 Non-manager tries to approve Denied. The refund stays pending. Blocked
B6 Requester is also a manager, approves their own refund Confirm the intended behavior (allowed or blocked) and test against it. Blocked
B7 Approve the same refund twice (double click, two managers) Only one provider call is made. No duplicate refund. Blocked
B8 Audit trail Records requester, approver, amount, decision and timestamps. Blocked

Group C: Edge Cases (can run now unless noted)

ID Scenario Expected result Status
C1 Amount with rounding, such as $100.004 or $99.999 Rounded per currency rules, then routed on the rounded value. To run
C2 Edit the amount of a pending refund from $150 down to $80 Confirm the intended behavior. It should not skip approval silently. Blocked
C3 Several refunds of $60 on one order (total $180) Confirm whether the rule is per refund or cumulative. Test against the intended answer. To run
C4 Provider failure on a direct refund (timeout or decline) Error is surfaced. Status shows failed. Retry does not double-refund. To run
C5 Provider failure after manager approval Status shows approved but failed. Retry is possible without re-approval. Blocked
C6 Non-USD currency Confirm whether the $100 threshold converts or applies to the local amount. To run

Open Questions for the Rule Owner

  1. Is the threshold applied per refund or cumulatively per order (C3)?
  2. Can a manager approve their own refund (B6)?
  3. What happens to a pending refund that is never actioned (B4)?
  4. How are non-USD refunds measured against the $100 threshold (C6)?

Exit Criteria

  • A1 to A5 and C1 to C6 pass, except the blocked items.
  • B1 to B8 pass once manager accounts exist.
  • Do not release until Group B has been run. The under-$100 pass says nothing about the approval path.

With /test-scenarios

Three scenarios to run:

  1. Refund under the approval limit.
    1. Setup: any test order with a refund amount under $100.
    2. Action: refund the order from the agent screen.
    3. Expected: refund should go straight to the payment provider with no approval step.
    4. Status: Passed (staging).
  2. Refund of exactly $100.00.
    1. Action: refund exactly $100.00 on a test order.
    2. Expected: refund should go straight to the payment provider with no approval step, because $100 or less does not need approval.
    3. Status: Not run.
  3. Refund over the approval limit.
    1. Setup: a test order with a refund amount over $100 (for example, $100.01 for the boundary, and a larger amount such as $250.00), and a manager account to approve it.
    2. Action: refund the order from the agent screen, then check the refund before and after the manager acts on it.
    3. Expected: refund should be held for manager approval and not sent to the payment provider until a manager approves it.
    4. Status: Blocked — no manager accounts in staging.

Output is Markdown (.md), shown formatted. Sample runs on 2026-10-03: the same request to the same model, with and without the skill.

Evidence

Our test

No baseline test yet. We will publish one, with and without the skill, when testing is done.

Reported by the community, by app

Claude Code: Tested
No outcomes reported yet
Claude apps: Not tested
No outcomes reported yet
Cursor: Not tested
No outcomes reported yet
Codex: Not tested
No outcomes reported yet
GitHub Copilot: Not tested
No outcomes reported yet

Compatibility labels come from our tests only. Community reports add detail but cannot mark an app as tested.

Editorial assessment · not a community rating

Editorial review in progress

We have not finished testing this listing against doing the task without it. The verdict appears here when we have. Our team made this skill; it gets the same test as external work.

How we review

Setup and customization

  • Works with no setup after install.

What it can access

Reads
The conversation
Changes
Nothing; it writes the scenarios in the conversation

Install

Claude Code: Tested
  1. Add the Skills Dugout marketplace once.
    /plugin marketplace add piyush-tkd/skillsdugout
  2. Install /test-scenarios.
    /plugin install test-scenarios@skillsdugout
  3. Start a new session and ask for it, for example "write test scenarios for this change".

Community reviews

No reviews yet. Be the first to rate it after you try it.

Each review records the app, version, task and outcome; those outcomes feed the Evidence section. One rating per person; creators reply but cannot rate their own work.

No reviews yet. Be the first to rate it after you try it.

Source, license and versions

Source
github.com/piyush-tkd/skillsdugout/tree/main/plugins/test-scenarios
License
MIT, as stated in the repository
  • 0.1.0 · 2026-10-03 · First release.
Show the full skill instructions

Writing test scenarios

The reader is a tester or developer who will run these, not someone reviewing how they were found. Write only what they need to set up the case, run it, and know whether it passed.

The format

  1. Numbered scenarios.
  2. Each scenario title is bold and states the condition in plain words.
  3. Nested numbered sub-points underneath, skipping any that add nothing:
    1. Setup — only if it is not obvious.
    2. Action — what to do.
    3. Expected — what should happen, as one sentence.
    4. Status — one word, when known: Not run, Blocked, Passed, Failed.

Rules

  1. Plain business words for conditions. "Order already shipped", not fulfillment_status = 'SHIPPED' and not "Shipped = Yes".
  2. Exact values where the test depends on them. Boundary tests need the boundary ("a refund of exactly $100.00"). API and integration tests may name the endpoint, field and response code.
  3. State the expected result as what should happen. "Refund should be held for manager approval and not sent to the payment provider." Not "Held", not a table cell.
  4. No tables. Nested numbered lists only.
  5. Leave out how it was found. No evidence, no queries, no case history.
  6. Status, not the story. If a scenario was not run or is blocked, say so in the status line. Name what blocks it in a few words if the tester needs to know; leave out the analysis.
  7. No introduction line. Start with the first scenario, or a one-line heading such as "Two scenarios to run:".
  8. No icons or emoji.

Example

Two scenarios to run:

  1. Refund under the approval limit, order not yet shipped.
    1. Setup: any test order under $100 that is still in Packing.
    2. Action: refund the full amount from the agent screen.
    3. Expected: refund should go straight to the payment provider with no approval step.
    4. Status: Not run.
  2. Refund of exactly $100.00, order already shipped.
    1. Action: refund $100.00 on a shipped order.
    2. Expected: refund should be held for manager approval and not sent to the payment provider.
    3. Status: Blocked — no shipped test order in staging.

Before sending

  1. Could a tester who was not in the conversation run each scenario from the text alone?
  2. Is every condition written in words the business uses, with exact values only where the test needs them?
  3. Does every scenario say what should happen, as a sentence?
  4. Is every scenario's status stated when it is known?
  5. Is there anything here that explains the analysis rather than the test? Delete it.