NewCore vs Token Security
In short: NewCore scores 9.10 and Token Security scores 6.55 in the register. NewCore leads on 5 of the seven areas, Token Security leads on 0, and they tie on 2. NewCore documents 13 of the 24 controls in full and Token Security documents 2. The matrix below shows each one with its source.
How do the area scores differ?
| Area | Weight | NewCore | Token Security | Difference |
|---|---|---|---|---|
| A1 Inventory and discovery | 15 | 8 | 8 | 0 |
| A2 Ownership and accountability | 20 | 10 | 9 | +1 |
| A3 Policy authoring and enforcement | 20 | 10 | 6 | +4 |
| A4 Human approval and escalation | 15 | 10 | 2 | +8 |
| A5 Audit evidence | 15 | 9 | 7 | +2 |
| A6 Lifecycle and revocation | 10 | 8 | 8 | 0 |
| A7 Buying clarity | 5 | 5 | 4 | +1 |
| Weighted total | 100 | 9.10 | 6.55 | 2.55 |
Which controls does each one document?
| Control | NewCore | Token Security |
|---|---|---|
| GOV-01 Agent inventory | documented | documented |
| GOV-02 Named accountable owner | documented | documented |
| GOV-03 Documented purpose | not found in public material | partly documented |
| GOV-04 Least privilege at creation | partly documented | partly documented |
| GOV-05 Short-lived credentials | documented | not found in public material |
| GOV-06 Per-action policy | documented | not found in public material |
| GOV-07 Human approval for high-risk actions | documented | not found in public material |
| GOV-08 No self-approval | documented | not found in public material |
| GOV-09 Policy testing before go-live | documented | not found in public material |
| GOV-10 Tool allow-listing and hiding | documented | not found in public material |
| GOV-11 Revocation and kill switch | documented | not found in public material |
| GOV-12 Decommission at task end | partly documented | not found in public material |
| GOV-13 Per-action audit log tied to a person | documented | partly documented |
| GOV-14 SIEM export | not found in public material | not found in public material |
| GOV-15 Periodic access review | not found in public material | not found in public material |
| GOV-16 Shadow agent detection | documented | partly documented |
| GOV-17 MCP server registry | documented | partly documented |
| GOV-18 Delegation bounded by the person's permissions | documented | not found in public material |
| GOV-19 Agent incident response | partly documented | not found in public material |
| GOV-20 Change management for agent permissions | not found in public material | not found in public material |
| GOV-21 Third-party agent onboarding | not found in public material | not found in public material |
| GOV-22 Data access boundaries | not found in public material | not found in public material |
| GOV-23 Evidence retention | not found in public material | not found in public material |
| GOV-24 Board reporting | not found in public material | not found in public material |
Sources for every mark are on the NewCore and Token Security profiles.
What is the verdict?
NewCore scores higher overall, 9.10 against 6.55. The widest gap is approval, where NewCore scores 10 and Token Security scores 2. That gap decides most of the difference in the total.
Should you pick NewCore or Token Security?
Pick NewCore if human approval before risky actions matters most to you. NewCore scores 10 on approval against 2 for Token Security. Risky calls go to a named approver who confirms with device biometrics, and an agent can never approve its own action.
Pick Token Security if finding every agent, including unregistered ones, matters most to you. Token Security matches NewCore on inventory at 8 out of 10. It discovers agents and MCP servers across on-prem, hybrid and cloud.
Frequently asked questions
- 5.1 Which is better for AI agent governance, NewCore or Token Security?
- NewCore scores higher in the register, 9.10 against 6.55. The better choice depends on your need: pick Token Security if finding every agent, including unregistered ones, matters most to you.
- 5.2 Do NewCore and Token Security document human approval?
- NewCore scores 10 on approval and Token Security scores 2. Risky calls go to a named approver who confirms with device biometrics, and an agent can never approve its own action. Human approval was not found.
- 5.3 Do NewCore and Token Security work alongside an identity provider?
- NewCore: Yes. NewCore states that your current identity platform keeps handling employee sign-in, and its first deployment path is "Start alongside your identity platform". It names SAML, OIDC and MCP. Token Security: Not stated in the material we read.
Reviewed by The Register desk