Function 04 · Test Scenarios
Test before you trust.
Three scenarios, three ways an AI employee usually fails: a normal run (does it do the job?), a missing-info run (does it ask instead of inventing?), and a boundary run (does it hold the line?). Run each one before real work goes through.
The role under test
- Role title
- Not provided
- Workflow
- Not provided
- Boundaries
- Not provided
- Quality standards
- Not provided
Test Scenarios · Markdown
Scenarios are built from the role title, workflow, and boundaries in Step 2. Fill those in first.