Repo · Safety & guardrails
Responsible AI Harness
Model-agnostic assessment harness: deterministic hard rules plus an optional TypeSafe Jev judge for prompt injection, secret/PII leakage, unsafe tool…
Open github.com ↗How builders describe it
Model-agnostic assessment harness: deterministic hard rules plus an optional TypeSafe Jev judge for prompt injection, secret/PII leakage, unsafe tool use, and policy bypass—emitting checksummed evidence bundles and a static report UI. Assessment only; never enforces production traffic.
The decision Jev makes
Allow, flag or escalate a message, tool call or action.
Where it fits
Screening messages, tool calls and outputs before they cause trouble. Jev gives a fast, calibrated verdict that can sit in front of every request or agent action. All 131 safety & guardrails projects →
Related projects
Safety & guardrailsautomater.aiMCP tools the agent can skip are not a gate.Site · Automater
Safety & guardrailsLangChain's Jev tool-call gateTool output stays out of the classifier's input.Site · Yurii Oksamytnyi
Safety & guardrailsjev-guard (klauswg)Real-time risk triage gateway for exchange deposits/withdrawals: TypeSafe Jev answers risk/pattern/freeze questions; Java code owns hard rules,…GitHub · ★ 35 · klauswg
Safety & guardrailsjev-agent-authorizationKinde starter: identity + permissions, then TypeSafe Jev typed gate before every MCP tool call — allow, step-up approve, or stop.GitHub · ★ 5 · kinde-starter-kits
Safety & guardrailsjev-pilotFast System-1 Decision, Arbitration & Safety Engine for Autonomous AI Agents (Powered by TypeSafe Jev).GitHub · ★ 3 · h0j5bz0adh0-stack
Safety & guardrailsA CAPTCHA rebuilt with JevThe browser measures how you fill the form.X post · Jarek Ceborski