- Home
- Skills
- Testing & Debugging
- ai qa uat tester
ai qa uat tester
Structured QA and UAT planning with evidence-based risk modeling and AI-specific validation.
Free
Works with the AI Agents you already use
ai qa uat tester
Example session with this skill installed
I need a P0 test suite for a new password reset flow. Users enter email, get a 6-digit code via SendGrid, and set a new password. It must handle invalid codes and expired links.
- Read your context and instructions
- Compiled the ai qa uat
I have decomposed the password reset requirements into a risk-prioritized test suite focused on P0 security and functional paths.
Password Reset P0 Test Suite
| ID | Priority | Scenario | Given/When/Then | Expected Result | Evidence to Capture |
|---|---|---|---|---|---|
| TC-001 | P0 | Happy Path Reset | Given valid email; When correct 6-digit code entered; Then allow password change. | User session is updated; old password rejected. | API 200 Response; DB hash change timestamp. |
| TC-002 | P0 | Expired Token | Given code > 15 mins old; When submitted; Then show "Link Expired" error. | Redirect to request new link; no password change. | UI Screenshot of error; SendGrid timestamp log. |
| TC-003 | P0 | Invalid Code | Given incorrect 6-digit code; When 3 failed attempts; Then lock request. | Account remains secure; rate limit header triggered. | 429 StatusCode or error message UI. |
| TC-004 | P0 | SQL Injection | Given ' OR 1=1 -- in email field; When submitted; Then fail gracefully. | No leakage of user data or bypass of auth. | Application error logs; WAF logs. |
UAT Sign-off Checklist
- Persona: Standard User
- Goal: Recover account access without support ticket.
- Success Criteria: User reaches dashboard with new credentials in under 2 minutes.
Next steps
- Provide the SendGrid integration logs to verify email delivery latency.
- Execute TC-003 to confirm rate-limiting is active in the staging environment.
- Confirm if the 6-digit code should be invalidated immediately after one failed attempt or after a specific threshold.
Connects securely to your tools. The creator never sees your data.
About this skill
The problem
Vague acceptance criteria and hand-wavy QA leading to production regressions. Developers often struggle to bridge the gap between "it works on my machine" and actual user acceptance testing.
What it does
- Decomposes requirements into structured Given/When/Then test cases with explicit evidence requirements.
- Builds risk-based P0-P3 test suites focusing on user impact and change surface.
- Conducts deep analysis of AI-specific features, including groundedness, prompt injection, and tool-call correctness.
- Audits execution evidence to generate release gate reports that separate observed facts from inferred risks.
- Standardizes defect reporting with reproducible steps, severity levels, and adjacent regression plans.
Why this beats prompting it yourself
General LLMs often hallucinate successful test results or give "looks good to me" feedback on AI outputs. This skill enforces a strict evidence-based contract that prevents "NOT RUN" tests from being marked as "PASS" and treats AI responses with rigorous groundedness checks.
Use cases
- Generating a P0/P1 regression suite for a critical production hotfix.
- Analyzing a set of mobile app screenshots to identify UI/UX defects against design specs.
- Designing UAT scenarios for multi-persona workflows in B2B SaaS applications.
- Validating AI agent behaviors for instruction adherence and safety refusals.
Known limitations
Cannot physically click UI elements or execute code; requires human or automated agent feedback as evidence. Cannot verify fixes without new evidence.
Cannot execute code, click UIs, or call APIs directly. Requires external logs, screenshots, or test evidence to verify results. Does not perform automated penetration testing.
How to use
- 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
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 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
3 uses
Times used to date
Free forever
No account required to browse
Trust & safety
Security scan
Security scan passed 2 months ago
- Free with an Agensi account
Permissions
No special permissions declared or detected