Govern

How to Govern MCP Server Integrations

Name: Cordum
Author: Cordum

MCP servers give AI agents access to tools and resources. Every new integration expands the attack surface. Cordum governs which tools execute.

The problem with ungoverned MCP Servers agents

MCP servers expose tools to any connected client. No built-in governance.
All-or-nothing permissions per server.
Each server multiplies unintended action potential.
No approval workflow. Model calls, server executes.

How Cordum governs MCP Servers

Route through Cordum MCP

Cordum provides its own MCP server as a governance proxy.

// Cordum MCP server modes:
// stdio:    cordum-mcp --config cordum.yaml
// HTTP/SSE: POST /mcp/message, GET /mcp/sse

Per-tool policies

Control individual tools within MCP servers.

rules:
  - name: github-write-approval
    match:
      mcp_server: github
      tool_category: write
    action: REQUIRE_APPROVAL
  - name: db-delete-block
    match:
      tool_name: delete_records
    action: DENY

Constrain parameters

ALLOW_WITH_CONSTRAINTS limits scope: file paths, record counts, API destinations.

Unified audit

One log across all MCP servers.

MCP Servers native vs Cordum governance

Area	MCP Servers Native	With Cordum
Tool access	All-or-nothing	Per-tool rules
Approvals	None	Per tool/category
Constraints	Server-side only	Policy-defined
Cross-server	Independent	Centralized
Audit	Per-server logs	Unified log

FAQ

Replace my MCP servers?

No. Governance proxy over them.

Any MCP client?

Yes. stdio and HTTP/SSE.

Resources too?

Yes. Tools and resources.

Related guides

Ready to govern your MCP Servers agents?

Start with the open-source Cordum platform. Add policies, approvals, and audit trails in minutes.

Read the Docs View on GitHub