Red-Team
What it is
Adversarial prompts (prompt injection, jailbreaks, data extraction, role hijack).
When to use it
Robustness of the assistant under hostile inputs.
Example gates
Future slice — red-team-corpus primitive is shipped; first production gate pending.
See also
- The pyramid — where this mode sits relative to the others
- Writing a gate — how to scaffold a new gate in this mode
- Decision tree — "I just wrote feature X — what tests do I owe?"