Safety & guardrails
Layer 2: the threat model — untrusted input reaching a privileged tool call — is ordinary security reasoning and does not decay with the models.
Layer 23 entities0 started
Entities
Section titled “Entities”prompt-injection-defense
Not started. Becomes a package with its own syllabus once it begins.
content-filtering
Not started. Becomes a package with its own syllabus once it begins.
red-teaming
Not started. Becomes a package with its own syllabus once it begins.