Feature
48 governance tools via Model Context Protocol. Drop-in integration with Claude Desktop, Cursor, and any MCP-compatible client.
Register tools with governance policies attached. Every tool call is evaluated against your rules before execution. Block dangerous flags, enforce parameter allowlists, and rate-limit tool usage.
Route tool calls through governance evaluation in real-time. Policies are checked locally in < 1ms, adding negligible latency. Supports allow, block, warn, redact, and escalate actions.
Route high-stakes tool calls to human reviewers directly through MCP. Approval requests appear in your MCP client with full context. Configurable timeout and auto-deny policies.
Every tool call generates an HMAC-signed compliance receipt. Full audit history accessible via MCP tools. Export logs for SOC 2, GDPR, and HIPAA compliance reporting.
Instantly stop any agent via MCP tool call. Freeze agents for investigation, resume when safe. Emergency shutdown propagates to all connected MCP clients within seconds.
Single governance server serves multiple MCP clients simultaneously. Shared policies, per-client configuration. Centralized audit log across all connected agents.
Add Tork governance to your MCP client config. Works with Claude Desktop, Cursor, Windsurf, and any MCP-compatible client.
claude_desktop_config.json
{
"mcpServers": {
"tork-governance": {
"command": "npx",
"args": ["@torknetwork/mcp-server"],
"env": {
"TORK_API_KEY": "tork_...",
"TORK_POLICY": "policy.yaml"
}
}
}
}$ npx @torknetwork/mcp-serverNo installation required — npx runs the latest version directly.
register_toolevaluate_inputevaluate_outputscan_piiredact_piicheck_policylist_policiescreate_policyupdate_policyapprove_requestdeny_requestlist_pendingget_audit_logexport_receiptsverify_receiptkill_agentfreeze_agentresume_agentget_agent_statuslist_agentsget_metricsconfigure_alertstest_policyget_scorehealth_checkOne line to install. 48 tools for governance. Zero latency overhead.