Silverfort MCP GatewaySilverfort | Datasheet DS-003 | Rev 2026-11-03 |
| Total 7.20 / 10 | Ranks third of 14 | |
| Status: Available as part of Silverfort AI agent security. | ||
In short: Silverfort scores 7.20 out of 10 and ranks third of 14 in this guide. Its strongest areas are per-call authorization and human approval, and its weakest are credential brokering and standards. Deployment is listed as SaaS. Pricing is quote-based, with no published price. Each score below has a reason and a dated source.
Features
- MCP clients redirected to a Silverfort gateway tenant URL
- Every tool call checked first; denied calls blocked, permitted calls forwarded and logged
- Verdicts: approve, block or step up to a human
- Each agent bound to a human owner
- Discovery of sanctioned and rogue agents across identity providers, clouds and SaaS
- Inline integrations with Google Cloud Agent Gateway and Microsoft Copilot Studio
- Immutable audit trail
Specifications
| Parameter | Value | Source |
|---|---|---|
| Product type | SaaS MCP gateway | [sf-ai] |
| Connection | Redirect to gateway tenant URL | [sf-ai] |
| Decision outcomes | approve / block / step up | [sf-ai] |
| Ownership | Agent bound to a human owner | [sf-ai] |
| Integrations | Google Cloud Agent Gateway, Microsoft Copilot Studio | [sf-google] |
| Standards named | MCP, OIDC | [sf-ai] |
| Price as published | Quote-based packages | [sf-pricing] |
How does Silverfort score on each criterion?
| Criterion | Weight | Score | Reason | Source |
|---|---|---|---|---|
| G1 Identity-bound per-call authorization | 25 | 9 | Every tool call hits the gateway first, and sessions are correlated to a person through your identity provider. | [sf-ai][sf-launch] |
| G2 Credential brokering | 20 | 5 | It removes standing privileges for agents using static API keys, but credential brokering is not described. | [sf-ai] |
| G3 Human approval and escalation | 15 | 9 | The verdict on a call can be approve, block or step up to a human. | [sf-ai] |
| G4 Server registry and discovery | 10 | 7 | It discovers sanctioned and rogue agents across identity providers, clouds and SaaS. | [sf-ai][sf-launch] |
| G5 Audit evidence | 10 | 9 | An immutable audit trail ties each call to the initiating person. | [sf-ai] |
| G6 Standards | 10 | 5 | MCP and OIDC are named; other MCP authorization standards were not found. | [sf-ai] |
| G7 Deployment and buying clarity | 10 | 5 | A SaaS gateway reached by redirect, with quote-based packages and no published price. | [sf-ai][sf-pricing] |
| Total | 100 | 7.20 | Sum of score times weight, divided by 100. | |
Typical deployment
SaaS gateway tenant; MCP clients are redirected to it.
Pricing as published
Quote-based packages; no published price.
Works alongside your identity provider?
Yes. Silverfort correlates agent sessions to a person through your identity provider. [sf-ai]
Frequently asked questions
- Does Silverfort work alongside my identity provider?
- Yes. Silverfort correlates agent sessions to a person through your identity provider.
- How is Silverfort deployed?
- SaaS gateway tenant; MCP clients are redirected to it.
- What does Silverfort cost?
- Quote-based packages; no published price.
How does it compare?
Revision history
| Rev | Change |
|---|---|
| 2026-11-03 | First issue of this datasheet. |
Sources
- Silverfort AI agent security [sf-ai] read 2026-11-03
- Silverfort unveils AI agent security [sf-launch] read 2026-11-03
- Silverfort pricing [sf-pricing] read 2026-11-03
- Silverfort and Google Cloud Agent Gateway [sf-google] read 2026-11-03
By The Datasheet desk · Published and reviewed 2026-11-03