Technology subcategory

Guardrails

Policy filters and safety layers that block or rewrite unsafe model/agent actions.

Guardrails enforce allow/deny before side effects: PII filters, topic blocks, tool firewalls.

They are complementary to human approval gates — automated policy for the common case, humans for judgment. See Evaluations, IAM, LLM Gateway.

See also

Technologies

Features

Stacks

Document management

GitHub in this term

Primary repositories linked from member packs and devices.