Repo · Safety & guardrails
jevshield
Python runtime security gate for agent tool calls: TypeSafe Jev Choice/Noul/Score dual-factor evaluation with fail-closed parsing,…
Open github.com ↗From the repository
Sub-100ms security gate for AI agent tool calls, powered by TypeSafe's Jev (System-1) decision model. Single-request Choice/Noul/Score evaluation, dual-factor blocking matrix, calibrated-confidence routing, fail-closed parsing, zero-config local fallback. LangChain-ready.
agent-securityai-agentsai-safetyguardrailsjevlangchainllm-securitymiddlewarepythonsecuritysystem-onetool-callingtypesafe
How builders describe it
Sub-100ms security gate for AI agent tool calls, powered by TypeSafe's Jev (System-1) decision model. Single-request Choice/Noul/Score evaluation, dual-factor blocking matrix, calibrated-confidence routing, fail-closed parsing, zero-config local fallback. LangChain-ready.
Python runtime security gate for agent tool calls: TypeSafe Jev Choice/Noul/Score dual-factor evaluation with fail-closed parsing, calibrated-confidence routing, and a zero-config local heuristic fallback when no API key is set. Distinct from Node jev-shield (MCP/Gateway firewall).
Framework-agnostic guard layer that puts Jev in front of an agent's tool calls: a Choice risk tier, a Noul destructiveness check and a Score blast radius decide allow, deny or ask.
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 229 safety & guardrails projects →
Related projects
Safety & guardrailssnapifRust crate and CLI that scores one agent tool call against a shipped policy and returns Auto, Review or Escalate, using TypeSafe Jev…GitHub · ★ 2 · snapif
Safety & guardrailsjev-sec.github.ioJev IDS website: one flow.GitHub · ★ 2 · jev-sec
Safety & guardrailsclaudetiquetteClaude Code plugin that rewrites a hostile prompt into a plain request.GitHub · ★ 2 · zahlekhan
Safety & guardrailsdsh-backdoor-guardGuard for DeepSeek Harness: every model-requested tool call is judged by a decision model, hosted TypeSafe Jev or a self-hosted LiquidAI d1, before…GitHub · miketako3
Safety & guardrailsopencode-auto-classifierOpenCode plugin that sends each shell command, file read, and edit to TypeSafe Jev for typed classifications, then folds those answers into…GitHub · schemaitat
Safety & guardrailspi-thunder-guardSafety guard for Pi: before each bash call (and writes outside the repo) TypeSafe Jev must be at least 85% sure the call keeps to a short restriction…GitHub · Mandalorian007