Press Kit
Resources and assets for media coverage of Cordum.
Company Boilerplate
Cordum is the policy and approval firewall for AI-agent actions. Start with Cordum Edge to govern Claude Code shell and file actions, expand to MCP Governance for built-in tool calls, and use the customer-managed Platform to centralize policy, approvals, workflows, and redacted evidence. Cordum's Safety Kernel evaluates configured policy before governed actions are dispatched.
Key Messages
“Stop AI agents before they take risky actions.”
Company tagline
“Policy-before-dispatch means each governed action entering Cordum is evaluated before execution, not after.”
Technical overview
“Cordum keeps the core boring so you can innovate at the edges.”
Philosophy
What is Cordum?
Cordum puts policy, approval, and redacted evidence in front of governed AI-agent actions before execution. Edge is the first product entry, MCP Governance expands the action surface, and the Platform is the customer-managed backend.
Core Features
- Safety Kernel — Policy-before-dispatch evaluation with ALLOW, DENY, REQUIRE_APPROVAL, or ALLOW_WITH_CONSTRAINTS decisions
- Approval Gates — Human review for guarded operations, bound to policy and action context
- Evidence Timeline — Redacted descriptors, hashes, approval references, and state transitions
- MCP Server — Built-in tools and Gateway transports with scoped discovery and approval-aware mutations
- Platform — Customer-managed workflows, policy, scheduling, and audit infrastructure
Quick Facts
- Start
- Cordum Edge
- Expand
- MCP Governance
- License
- BUSL-1.1
- Platform model
- Customer-managed
