write-test-scenarios
Skill4 starsChanged 4 months agoMITView on GitHub
Generate manual test scenarios (UC/TC) from the spec file after implementation — use cases and step-by-step test cases for human testers to verify the feature by hand. Output saved to .ai/tasks/<name>-manual-tests.md.
In this repository12 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- .github/copilot-instructions.mdCopilot instructions
- clarify-scopeSkill
- create-pr-descriptionSkill
- debugSkill
- implement-tddSkill
- plan-tasksSkill
- review-prSkill
- setup-ai-contextSkill
- shipSkill
- sync-ai-contextSkill
- write-specSkill
The file
We split this skill into the rules it gives an agent
4 rules. About 0 are common, 2 are distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈2
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 44
Create .ai/tasks/<task-name>-manual-tests.md using the following format:
- Very specific: line 132
Verify each item in ## Test plan of the spec has at least 1 TC:
- Distinctive: line 140
If any Test plan item has no TC → add it before saving.
- Distinctive: line 144
After creating the file, report:
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
Write Test Scenarios
Generate manual test documentation (UC + TC) based on the written spec, so the user can verify the feature by hand after implementation is complete.
Output: file .ai/tasks/<task-name>-manual-tests.md
Run
Step 1 — Read inputs
Read: .ai/tasks/<task-name>.md ← get Goal + Test plan + Constraints
Read: .ai/context/domain-glossary.md ← use correct domain terminologyList all items in ## Test plan — this is the list of UCs to write.
Step 2 — Identify Use Cases
Group test plan items into UCs by user flow. One UC = one user journey with a clear start and end.
Example:
Test plan items:
- happy path: password reset succeeds
- edge case: token expired
- edge case: token reused
- no regression: login still works normally
→ UC-01: Successful password reset
→ UC-02: Handle invalid token
→ UC-03: Login flow unaffected (regression)Step 3 — Write test file
Create .ai/tasks/<task-name>-manual-tests.md using the following format:
# Manual Tests: <task-name>
**Feature:** <short description from spec Goal>
**Tester:** ___________
**Test date:** ___________
**Environment:** ___________ (staging / local / production)
---
## UC-01: <use case name>
> <1-2 sentence description of the use case — who does what, and why>
**Preconditions:**
- <condition that must be true before starting the test>
- ...
---
### TC-01-01: <specific test case name>
| | |
|---|---|
| **Priority** | High / Medium / Low |
| **Type** | Happy path / Edge case / Regression / Error handling |
**Steps:**
1. <specific step — detailed enough for someone without code knowledge to follow>
2. ...
3. ...
**Expected result:**
- <what must happen — specific and observable>
- ...
**Actual result:** ___________
**Pass / Fail:** ⬜ Pass ⬜ Fail
**Notes:** ___________
---
### TC-01-02: <next test case in the same UC>
...
---
## UC-02: ...Step 4 — Rules for writing good TCs
Steps must be specific enough for someone without code knowledge to follow:
# Good
1. Open browser, navigate to http://localhost:3000/forgot-password
2. Enter "test@example.com" in the "Email" field
3. Click the "Send password reset link" button
4. Check the inbox of test@example.com
# Bad
1. Call the reset password API
2. Check the resultExpected result must be observable:
# Good
- Page shows message "Email sent. Please check your inbox."
- Email arrives within 60 seconds with subject "Reset your password"
- Link in email matches the format https://.../reset?token=...
# Bad
- System processes correctly
- No errorsPreconditions must be clear:
- What data must exist (user account, order, ...)
- Required system state (email server running, feature flag enabled, ...)
- What permissions the tester needs
Step 5 — Ensure coverage
Verify each item in ## Test plan of the spec has at least 1 TC:
[ ] happy path → TC-XX-XX
[ ] edge case 1 → TC-XX-XX
[ ] no regression → TC-XX-XX (regression)If any Test plan item has no TC → add it before saving.
Step 6 — Report
After creating the file, report:
Created: .ai/tasks/<task-name>-manual-tests.md
- X use cases
- Y test cases (Z high priority)Quick format guide by TC type
Happy path: Primary actor, normal conditions, straight-through flow from start to finish.
Edge case: Boundary inputs (empty, too long, special characters), abnormal states (expired, already used, duplicate), concurrent actions.
Error handling: System responds correctly when an error occurs — clear message, no crash, no sensitive information leaked.
Regression: Existing flows not directly related to the new feature — must verify they still work normally.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

