Read change set
Separate touched code, current tests and coverage gaps.
Context TriageQuality Engineering Demo
A test agent that plans cases from a diff, generates runnable tests, critic-checks stability, and records progress.

Clickable pattern simulator
This simulator does not replace the CLI run, but it makes each step, state, log, gate and final output visible in the browser.
Separate touched code, current tests and coverage gaps.
Context TriageTurn the diff into boundary, regression and negative cases.
Plan-and-ExecuteCreate pytest-style test bodies from the plan.
Tool DispatchCheck determinism, assertions and fixture style.
Generator-CriticRepair once when tests fail for mechanical reasons.
Self-Heal LoopRecord generated cases to avoid duplicated work.
Progress Tracking
| Pattern | Design role |
|---|---|
| Plan-and-Execute | Generation needs an explicit test plan before code emission. |
| Generator-Critic | Generated tests must be checked by a separate role. |
| Self-Heal Loop | A bounded repair loop handles mechanical failures without hiding ambiguity. |
| Progress Tracking | The agent should remember generated coverage across PRs. |
A test that looks good but cannot run is negative value.
The best critic is grounded in deterministic checks: import, assertion, fixture, and test result.
Without progress tracking, the same cases get regenerated again and again.
cd workshops/workshop2-test-generation
python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt
python src/main.py
open visual/index.htmlworkshops/workshop2-test-generation/README.mdimages/training/visual/workshop2-test-generation-blueprint.png