What it is
Auditable auto mode for agent tool calls.
What it does: Before a risky tool runs, Jev scores destructive/exfil/privilege/secret/off-task risks and toolgate maps them to allow/ask/deny.
Why it belongs in the showcase: Strong star signal and a measured, multi-surface firewall on hosted Jev.
Caveats: Operator policy ownership; multi-endpoint routing.
Evidence
Public MIT repo, created 2026-09-18, 29 stars as of 2026-11-04. README documents npm @riskaverse/toolgate, seven risk questions plus authorized mitigator, Jev via TypeSafe/OpenRouter/Vercel/OpenCode Zen, and AI LABS feature coverage.
Caveats
- Policy YAML and thresholds are operator-owned
- Can route through non-TypeSafe System One-compatible endpoints
- Evaluation claims are author/third-party reported
Unofficial case page. Stars / list position observed for heat; do not treat heat as a quality SLA. Review the upstream README before you depend on it.