Add Cordum — production safety layer for AI agent orchestration #696
Description
Project
- Name: Cordum
- Repository: github.com/cordum-io/cordum
- Suggested Category: Privacy and Security / Serving Infrastructure
Description
Cordum is a safety-first agent orchestration platform designed for production AI/ML agent deployments:
- Safety kernel — gRPC policy evaluation before agent tool execution (deny/escalate/allow)
- Output scanning — PII, secrets, prompt injection detection on agent output
- Job scheduler — model-aware routing, topic-to-pool dispatch, distributed locks, dead-letter queue
- Workflow engine — DAG execution with approval gates, fan-out, conditional branching
- Audit trail — every policy decision and state transition recorded
- Horizontal scaling — 2-6 replicas per service, production-ready
7 Go services, React dashboard, Docker + K8s manifests. Built for production multi-agent environments.