TRUSTIVAN NHI Security
The identities no person signs in as
Service accounts, workload identities, API keys and CI/CD robots outnumber the people in most estates, and few of them have an owner anyone could name. TRUSTIVAN NHI Security inventories the ones its connectors can see, says who owns each and what it can reach, and decides whether an AI agent's action may proceed.
One product, two parts
AppSec and NHI Security, one sign-in
TRUSTIVAN AppSec governs what your software contains. TRUSTIVAN NHI Security governs the identities your software runs as. They share one sign-in, one organization model and one address; each keeps its own backend and database.
TRUSTIVAN AppSec
Vulnerability, SBOM, container and repository scanning, and the finding lifecycle, policy and verdicts that follow.
TRUSTIVAN NHI Security
Non-human identity inventory, ownership, reach, findings and runtime authorization for AI agents. This page.
What it does
From an inventory to a decision
Each stage below is built and tested in the product today.
For a plain-language walkthrough of why non-human identity needs a different discipline than human access review, readWhat is non-human identity (NHI) security?
Discover
Connectors for AWS, GitHub, Kubernetes and HashiCorp Vault inventory the non-human identities each provider reports — IAM roles and machine users, app installations and bot accounts, Kubernetes service accounts, Vault machine entities — and leave people's accounts out. An MCP connector lists the tools a Model Context Protocol server exposes.
Attribute and map
Each discovered record becomes one identity with an owner, and a graph links identity to credential to permission to resource — so a question like “what can this service account actually reach, and how” has an answer.
Assess
Policies run over that evidence and produce findings and derived risk per identity. An exception is scoped, approved and expires; it never silently widens.
Authorize at runtime
When an AI agent in the inventory tries to act, a runtime gate decides allow, challenge or deny against declared tool entitlements and runtime rules, and records the decision immutably.
Control
Just-in-time access grants bounded at seven days, where nobody approves their own request; agent runtime credentials that are shown once and stored as a digest; reversible quarantine of an identity.
Audit
An append-only, hash-chained trail per tenant, with tenant isolation enforced by row-level security in the database rather than by application code.
Connectors
Five, each asking for the least access it needs
A connector stores a reference to its credentials, never the credentials. It classifies an identity from what the provider reports about it, never from its name, and reports an unrecognised kind as unknown rather than guessing.
- AWS
- IAM roles and IAM users that are evidently machine identities, with the relationships between them — and, alone among the connectors, access-key metadata and inline-policy permissions.
- GitHub
- App installations in an organization, and the accounts GitHub itself marks as bots — read with two read-only permissions. Automation running under an ordinary user account is not guessed at from its name, so it is not listed.
- Kubernetes
- ServiceAccounts, which are non-human by construction; the role bindings that say what each can reach; and the workloads that run as it. Secrets are never read.
- HashiCorp Vault
- Entities that arrive through machine auth methods; human-capable ones are excluded.
- MCP tool servers
- The tools a server publishes. Implemented and tested against a deterministic fake server; not yet run against a live one.
Discovery runs when someone synchronises a connector. A scheduler for unattended collection exists and is off by default.
Scope, stated plainly
What NHI Security does not do
An evaluation that finds these later has wasted your time, so they are here first.
Not available
- It does not discover AI agents. No connector reports one and no screen creates one. The runtime gate authorizes the agents an estate's inventory holds; it cannot find ones nobody recorded.
- It has not been deployed to production.It is built and tested, and runs locally and in containers.
- It changes nothing in your cloud by default.Remediation is proposed, approved and dry-run inside the product; executing it against a provider is off unless an operator turns it on.
- No behavioural analytics beyond two narrow runtime rules, no agent-to-agent lineage, and no endpoint telemetry.
- An agent's stated intent is recorded for the investigator; no rule decides on it.
- Certification, delegation, drift and compliance are API surfaces without console screens yet.
Why say so
NHI Security refuses to assert what it cannot evidence. A deployment that has connected nothing shows a dash and the reason, never a confident zero; a missing last-used date stays missing rather than becoming “unused”. The same rule applies to this page.
One account, both parts
Sign in to TRUSTIVAN once. NHI Security opens from the same dashboard, in the organization you are already working in.