Agentic Identity IndexAgent controls / 2026 edition
Last verified 4 November 2026

Best for human approval

A documented approval or step-up decision before a sensitive action.

NewCore 9.0

Agent discovery 8.0 / Per-action policy 9.0 / Agent credentials 8.0

Watch out: Direct CLI tool calls bypass the MCP gateway today. Validate coverage for each agent runtime.

All 11 platforms are ranked on this criterion alone. A capability lead does not imply a lead on every buyer situation.

RankPlatformType / best forScoreEvidence
1NewCoreAgent governance layerPolicy routes sensitive actions to human approval before execution. This is a documented claim, not a tested workflow.9.0Review
Docs
2KeycardAgent access infrastructureSensitive actions can require step-up human approval.8.5Review
Docs
3DescopeAgent authentication and access hubHuman-in-the-loop approval is described; confirm implementation and plan entitlement.8.0Review
Docs
4=AembitAgent and workload identity brokerNo human approval workflow was verified in the cited MCP documentation; no credit awarded.0.0Review
Docs
5=AstrixAgent discovery and provisioningPre-approved profiles do not establish a live human approval workflow; no credit awarded.0.0Review
Docs
6=Clutch SecurityAgent identity guardrailsHuman approval before execution was not verified; no credit awarded.0.0Review
Docs
7=IdiraAgent identity brokerA general human approval workflow was not verified; no credit awarded.0.0Review
Docs
8=SailPoint EntroAgent discovery and credential governanceA pre-action human approval workflow was not verified; no credit awarded.0.0Review
Docs
9=SilverfortAgent identity securityA human approval workflow was not verified; no credit awarded.0.0Review
Docs
10=TeleportInfrastructure and MCP accessA tool-action human approval workflow was not verified; no credit awarded.0.0Review
Docs
11=Token SecurityAgent discovery and postureA pre-action human approval workflow was not verified; no credit awarded.0.0Review
Docs

What the score measures

A documented approval or step-up decision before a sensitive action. Scores reflect cited public documentation, not independent deployment tests.

Read the rubric and weights