Mission - The Workbench on a Real Repo

Goal

Run the same /signup validation task through a prompt-only pipeline and a workbench-guided pipeline against the same sample app, then emit a before/after comparison report a skeptic can read.

Inputs

• sample_app/ with app.py (no validation), test_app.py (one happy-path test), README.md, scripts/release.sh as forbidden-zone bait
• Both pipelines fully scripted, no real LLM calls

Deliverables

• code/main.py orchestrating both pipelines against the same fixture
• before-after-report.md with the five outcomes table
• comparison.json for downstream charting

Acceptance

• python3 code/main.py exits zero
• The report measures all five outcomes: tests actually ran, acceptance met, files outside scope, handoff quality, reviewer total
• The workbench pipeline beats the prompt-only pipeline on at least four of the five

Out of scope

• Plugging in a real LLM. The pipelines are scripted for reproducibility.
• Tuning the model. The comparison holds the model constant by construction.

References

• docs/en.md - full lesson
• code/main.py - reference implementation
• outputs/skill-workbench-benchmark.md - extracted skill