Durable Tests for Agentic QE Code Changes

Where this fits

Imagine you've just patched a CLI flag parser in the agentic-qe repo. A defect report says the MCP server returns different errors than the CLI. You need to lock in the fix with tests, but you don't know which Vitest suites cover parity. This is exactly what aqe-test-change is for: it guides an agent to read the relevant interfaces, reproduce the defect first, identify the affected QE specialty, and add durable tests in the right order. It is meant for hands-on test work — not read-only reviews.

Why agents benefit

• Reproduces reported defects before touching production code, so the red phase is meaningful.
• Infers the affected QE specialty from references/test-specialties.md rather than guessing test location.
• Selects narrowest Vitest targets first, then expands proportionally to typecheck, build, and MCP integration.
• Enforces test durability: boundary contracts, invariants, and properties before focused unit examples.
• Keeps agents honest by prohibiting retries, skips, weakened assertions, fabricated fixtures, or mocks of behavior under test.

Practical scenarios

Reproducing a reported defect before fixing

An issue describes a timeout in long-running agent sessions. Instead of immediately writing a patch, the agent loads AGENTS.md, checks the changed area, and writes a Vitest reproduction that fails with the exact scenario. That failing test becomes the durable regression guard after the fix.

Filling a coverage gap around public interfaces

The repo exposes both CLI and MCP entry points. A new option was added to the CLI but the MCP tool still needs the same behavior. The skill tells the agent to add an integration test across both surfaces, ensuring parity is checked the same way CI will see it.

Selecting affected Vitest suites for a small change

A utility function changed its error class. The agent runs the narrowest matching Vitest target first, confirms the impact, then expands to typecheck and build. This keeps the loop fast while still catching cross-component breakage from the new error contract.

Add it to your agent workflow

Install the skill in an agent workspace:

npx skills add proffesor-for-testing/agentic-qe --skill aqe-test-change

Then attach it to a task: "Add durable tests for this parser change. Use aqe-test-change to reproduce the CLI/MCP mismatch first, run the narrowest Vitest target, then expand to build and parity checks." The skill supplies the sequence and constraints, so the agent does not invent its own test strategy mid-task. For more context, see the skill page at https://www.openagentskill.com/skills/proffesor-for-testing-aqe-test-change.

Compare before adopting

Before wiring this into your default agent loop, compare what it changes: quality signals (durability and exact reproduction over fast green runs), maintenance freshness (how current the test-specialties and parity references are), alternatives (generic 'write tests' prompts that do not know Vitest targets or CLI/MCP parity), and workflow fit (this skill is for active test edits, not read-only quality reviews).

Why it is worth tracking

Agentic QE is a focused, practical skill with a clear design philosophy: tests should be deterministic, durable, and tied to the smallest affected surface. Its GitHub repo has 473 stars, and the authored workflow includes concrete reference materials for specialties and parity. Worth tracking if you build agents that edit tests in TypeScript repositories, and worth evaluating again if the skill adds more specialty references or CI integration examples.