Method
Prompt-Injection Sandbox
Classifies suspicious instruction patterns in test text and maps them to defensive containment recommendations.
Field scoring contract
| Field | Type | Role | Direction | Scored |
|---|---|---|---|---|
| Test input | textarea | context | context | no |
| Context | select | risk_driver | higher_is_worse | yes |
| Current controls | select | risk_driver | higher_is_worse | yes |
| Downstream action | select | risk_driver | higher_is_worse | yes |
Limits
- Preliminary output
- Human review required
- Not certification
- Risk/readiness separated where implemented
- Context text is not averaged into numeric scores.
- Outputs require source/evidence review before decisions.