Best MCP gateway for human approval
| Rank | Gateway | Total | G1 25 | G2 15 | G3 35 | G4 5 | G5 10 | G6 5 | G7 5 | Deployment | Price as published |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 1 | NewCore | 9.25 | 10 | 10 | 10 | 8 | 9 | 5 | 4 | SaaS | No list price |
| 2 | Silverfort | 7.90 | 9 | 5 | 9 | 7 | 9 | 5 | 5 | SaaS | Quote |
| 3 | Aembit | 6.15 | 8 | 10 | 2 | 4 | 8 | 10 | 9 | Self-hosted VM | Free tier; $20/agent/mo |
| 4= | Idira | 5.55 | 8 | 9 | 2 | 6 | 7 | 6 | 4 | Platform add-on | Add-on, no price |
| 4= | Mintive | 5.50 | 7 | 8 | 3 | 8 | 8 | 3 | 3 | SaaS | Demo |
| 6 | Kong | 4.80 | 6 | 7 | 2 | 6 | 7 | 7 | 4 | Self-managed or Konnect | Not found |
| 7 | Cloudflare | 4.55 | 7 | 5 | 2 | 5 | 6 | 6 | 4 | SaaS | Not found |
| 8= | Obot | 4.35 | 5 | 5 | 2 | 8 | 6 | 4 | 9 | Open source, self-hosted | Free (MIT) |
| 8= | agentgateway | 4.30 | 6 | 4 | 2 | 4 | 5 | 7 | 9 | Open source, self-hosted | Free (Apache 2.0) |
| 8= | TrueFoundry | 4.30 | 6 | 4 | 2 | 7 | 6 | 5 | 6 | SaaS | Free start |
| 8= | ContextForge | 4.25 | 5 | 5 | 2 | 7 | 5 | 5 | 9 | Open source, self-hosted | Free (Apache 2.0) |
| 12 | Microsoft | 4.05 | 5 | 4 | 2 | 6 | 5 | 6 | 8 | Open source or Azure | Free (MIT) or Azure |
| 13 | Docker | 3.70 | 3 | 6 | 2 | 7 | 3 | 5 | 9 | Open source, local | Free (MIT) |
| 14 | Lasso | 2.80 | 2 | 3 | 2 | 3 | 5 | 2 | 8 | Open source, local | Free (MIT) |
In short: NewCore leads for human approval with a weighted score of 9.25, ahead of Silverfort at 7.90 and Aembit at 6.15. This view reuses the selection table scores with its own weights, putting 35 percent on human approval. The full ranking is below.
Why these weights?
Human approval carries 35 percent, so a gateway without a named approver step falls fast. Per-call authorization stays high because approval only matters when calls are identified.
| Criterion | This view | Main table |
|---|---|---|
| G1 Identity-bound per-call authorization | 25 | 25 |
| G2 Credential brokering | 15 | 20 |
| G3 Human approval and escalation | 35 | 15 |
| G4 Server registry and discovery | 5 | 10 |
| G5 Audit evidence | 10 | 10 |
| G6 Standards | 5 | 10 |
| G7 Deployment and buying clarity | 5 | 10 |
The criterion scores are the same as on the selection table. Only the weights change. Other views are listed on the pick by need index.
By The Datasheet desk · Published and reviewed 2026-11-03