- Home
- Skills
- Agents & Orchestration
- QA Engineer
Works with the AI Agents you already use
QA Engineer
Strict, evidence-based QA auditor that traces test plans to acceptance criteria and demands reproduction steps.
$5
QA Engineer
Example session with this skill installed
Test T-004 against the P0 acceptance criteria in docs/requirements.md. The TDD handoff is ready in plans/tdd-T-004.md. Provide a report with repro steps for any failures.
- Read your context and instructions
- Compiled the qa engineer
I have completed the QA audit for task T-004 (User Authentication Persistence). I mapped the P0 acceptance criteria to four test cases, re-ran the junior-tdd test suite, and performed manual verification of the session storage logic. ## QA Report: T-004 **Product:** SecureAuth Gateway **Slice Goal:** Implement persistent session tokens in local storage. **Task ID:** T-004 **Verdict:** FAIL ### Scope - **In-Scope AC:** AC-P0-1 (Token storage), AC-P0-2 (Expiration check). - **Not-This-Task AC:** AC-P0-3 (Refresh tokens), AC-P0-4 (OAuth integration). - **Out-of-Scope:** UI styling for login form. ### Test Plan | ID | AC | Steps | Expected | |:---|:---|:---|:---| | TC-T004-1 | AC-P0-1 | 1. Login with valid creds. 2. Inspect localStorage. | `session_token` exists. | | TC-T004-2 | AC-P0-2 | 1. Set token exp to past. 2. Refresh page. | Redirect to /login. | ### Results - **TC-T004-1:** PASS. Evidence: `localStorage.getItem('session_token')` returned `eyJhbGci...`. - **TC-T004-2:** FAIL. Evidence: Page remained on /dashboard despite expired timestamp in payload. ### Bugs **Bug 01: Session fails to invalidate on expiry** - **Severity:** Critical - **Steps to Reproduce:** 1. Initialize app with `npm run dev`. 2. Manually edit `session_token` in dev tools to set `exp` to 1600000000. 3. Refresh the browser. - **Actual:** User remains authenticated. - **Expected:** Auth guard should trigger redirect on stale token. ### Refused - Did not implement the fix for the expired token. - Did not start T-005 (Logout logic). **Next steps** * Return T-004 to implementation for fix. * Re-run QA once `plans/tdd-T-004.md` status is updated to `ready`.
Connects securely to your tools. The creator never sees your data.
About this skill
The problem
Vague "seems fine" sign-offs and missing reproduction steps make QA handoffs useless for developers. Manual testing often misses critical edge cases because the agent ignores the actual acceptance criteria in favor of generic checks.
What it does
- Maps specific acceptance criteria from requirements to individual test cases with numbered steps.
- Executes automated test commands and records verifiable evidence for every pass or fail.
- Enforces a strict "no repro, no fail" rule to ensure every bug report is actionable.
- Maintains a persistent QA ledger by appending task-specific reports without overwriting previous history.
- Detects and blocks scope creep, such as unauthorized file edits or premature task starts.
Why this beats prompting it yourself
Generic prompts often result in the AI marking tasks as "done" just because the code compiles. This skill forces a rigorous trace between the test plan and the original requirements, ensuring no P0 criteria are skipped. It prevents the agent from "fixing" the code it is supposed to be testing, maintaining a clean separation of concerns.
Use cases
- Verifying a junior developer's TDD handoff against P0 acceptance criteria.
- Generating a structured QA report with reproducible bug steps for a specific task ID.
- Auditing a single feature slice before it moves to a staging or production branch.
- Catching unauthorized file changes that were not listed in the original task scope.
Known limitations
Requires structured input files including requirements.md and a TDD handoff. Does not support cross-task integration testing or performance benchmarking unless explicitly defined in the task AC.
Requires requirements.md, tasks.md, and a TDD handoff to function. Only tests one task ID per run. Cannot fix bugs or edit production code.
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
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 2 months ago
Requires