Continuous Non-Human Identity (NHI) verification

Non-Human Identities (NHIs) — service accounts, AI agents, MCP tool clients, CI/CD runners — often receive long-lived credentials with no continuous re-validation. Control Core verifies NHI context on every request at the Bouncer and surfaces them in Agent Access Controls for operators to profile and govern.

Control Core is not an identity directory. Credential lifecycle, enrollment, and secret rotation stay with your existing IAM or NHI management systems. Any vendor can push attributes through a single generic PIP API.

What gets verified

Signal	Source	Control use
`input.subject.nhi.id`	SPIFFE ID (XFCC), JWT `sub` / `client_id` / `azp`, or `x-cc-agent-id`	Portable subject key
`input.subject.nhi.source`	Resolution trust tier: `spiffe` \| `jwt` \| `header`	Weight enforcement by trust
`input.subject.nhi.identity_type_hint`	Derived from resolution path	Classify workload class
`data.pip.nhi_attributes[subject_key]`	Control Plane NHI registry + `POST /v1/pip/subject-attributes`	Privilege score, scopes, status

Lifecycle in Agent Access Controls

Status	Meaning
observed	First seen in Bouncer traffic — Discovery Inbox
profiled	Operator set identity type, scopes, privilege score
governed	Access control bound; Rego templates enforce

Open Observe → Agent Access Controls (/observe/nhi) in the Control Plane.

How to use

Deploy a Bouncer on agent/API paths (deployment modes) (~15 min).
Allow traffic — first-seen NHIs auto-register as observed (disable with NHI_AUTO_DISCOVERY=false if needed).
Profile NHIs — set identity type, declared scopes, privilege score; mark profiled or governed.
Optional — sync from your IAM / NHI directory — any system posts portable attributes (no vendor SDK):

curl -sS -X POST "$CONTROL_PLANE/v1/pip/subject-attributes" \
  -H "Authorization: Bearer $TOKEN" \
  -H "Content-Type: application/json" \
  -d '{
    "subject_key": "spiffe://example.internal/ns/prod/sa/billing-agent",
    "attributes": [
      {"name": "privilege_score", "value": 3},
      {"name": "identity_type", "value": "orchestrator_agent"},
      {"name": "declared_scopes", "value": ["read:invoices", "tools/read"]}
    ]
  }'

Enable NHI templates — NHI continuous verification, NHI privilege scope, NHI identity type gate under Controls → Templates → AI Security.
Optional M2M tokens — POST /auth/nhi-token with grant_type=client_credentials issues a short-lived JWT (subject_type=nhi) for registered NHIs.

Example Rego (shipped templates read PIP data):

nhi_id := object.get(object.get(input.subject, "nhi", {}), "id", "")
nhi_attrs := object.get(data.pip.nhi_attributes, nhi_id, {})

deny if {
  nhi_id != ""
  object.get(nhi_attrs, "token_freshness_ms", 0) > 300000
}

Troubleshooting: Decisions lack input.subject.nhi?

Confirm SPIFFE XFCC, JWT with subject_type=nhi / client_id, or x-cc-agent-id is present

Check Audit Logs for actor_type=ai_agent

Full reference: Troubleshooting

Troubleshooting: PIP attributes missing in Rego?

Confirm the NHI attributes data source is healthy (Policy Bridge Completed)

Re-save the NHI profile or re-post /v1/pip/subject-attributes

See PIP getting started

How to review / audit

Agent Access Controls — Discovery Inbox + registry by status
Audit Logs — filter agent_id_str / actor_type=ai_agent
Grafana dashboard cc-02-nhi-activity when configured (see Grafana dashboards)

Periodic review checklist:

New observed NHIs reviewed within your SLA
Production NHIs are governed with declared scopes
External directory attributes stay in sync via /v1/pip/subject-attributes

Next steps: MCP blast radius · Agent framework connectors · AI governance overview