Know every identity.Own every one.Explain every risk.
Trustivan discovers service, workload and integration identities in AWS, GitHub, Kubernetes and HashiCorp Vault, classifies each one, attaches an accountable owner and a lifecycle, and explains its risk with the evidence attached.
Available today: connector contract, identity discovery, classification, ownership, lifecycle, identity graph, risk context with evidence
Inventory · 9 identities · 3 owners · 2 unowned
Sample environment
PTPlatform teamowner
DSData servicesowner
MLML platformowner
svc-billing-exportService
gha-deployAutomation
payments-saWorkload
vault-approle-etlService
analytics-gh-appIntegration
no owneredge-gateway-saWorkload
research-agentAI agent
finance-toolsMCP server
dormant 140dlegacy-export-keyService
Service
Workload
Integration
Automation
AI agent
Unowned
Illustrative identity population · sample environment · agents come from the demo seeder, not discovery
The problem
Sprawl. Orphans. The wrong owner.
Machine identities outnumber the people who created them — and most were never recorded anywhere a security team can see.
Identity sprawl
Every cloud account, cluster, pipeline and SaaS tenant mints its own identities, in its own format, with no shared inventory
Orphaned identities
Creators leave and projects end, but the service accounts and keys they made keep authenticating with nobody accountable
Ownership mismatch
The team on the tag is not the team that runs the workload, so attestation, rotation and offboarding land on the wrong desk
A list of accounts does not fix this. A relationship does: who owns this identity, what it holds, what it can reach and what it does — kept current by discovery, not by a spreadsheet.
How it works
Discover. Classify. Own. Explain.
Four capabilities on one identity graph, each producing evidence the next one relies on — available today.
Discover
Discover every identity
Connectors inventory identities where they live — AWS IAM, GitHub, Kubernetes and HashiCorp Vault, plus the tools on MCP servers — and keep that inventory current through a sync lifecycle you can audit.
Identity classes that exist: IAM roles and machine users, access keys, Kubernetes service accounts, GitHub Apps and bots, Vault AppRoles and entities, and MCP tools.
Sync you can audit: each connector run records what it read, what changed and when, so an inventory gap is visible rather than silent.
Search across everything: filter by type, lifecycle, provider or ownership confidence, and sort by last use, from one place.
Illustrative · sample environment · agents are not discovered
Illustrative · sample environment · agents are not discovered
Classify
Classify and contextualize
Type comes from the provider, ownership carries its confidence and reach comes from the identity graph — the account it lives in, the roles it can assume, the resources it reaches. Every record names its source, and an inferred owner is labelled inferred until a person assigns one.
Taxonomy that matches reality: IAM roles, service accounts, GitHub Apps and bots, Vault AppRoles and MCP servers, typed from what the provider reports.
Context that compounds: classification feeds ownership, risk factors and effective access on the same graph.
Unknown is a valid value: a field without evidence shows unknown instead of a confident guess.
Non-human identity · discovered 14 months ago · AWS IAM connector
Service identityHigh criticalityOwner unknown
Classification
Type
Service identity · IAM user with access keyAWS IAM
Presence
Present in the last complete syncsync run
Scope
AWS account prod-coreconnector scope
Reach
High impact — reaches billing resourcesidentity graph · resource impact
Lifecycle
Created 611 days ago · key never rotatedIAM snapshot
Context
Owner
Unknown — no Owner tag, none assignedAWS IAM tags
Suggested owner
Platform team (Team tag)AWS tag · inferred
Credential
1 access key · rotation overdueIAM snapshot
Last used
Unknown — not reported for this identityAWS IAM
Can reach
2 roles · 3 high-impact resourcesidentity graph
Own
Own every identity and its lifecycle
Every identity gets an accountable owner and a lifecycle. Owners come from provider tags or are assigned by a person, pointed at teams wherever possible so accountability outlasts any one engineer — and the identities nobody owns become findings.
Confidence, not assumption: a tag hint stays inferred until a person assigns the owner, and the change is recorded with a date.
Dormancy surfaced early: identities that still hold an active credential but, where the provider reports last use, no longer authenticate are flagged.
Ownership that survives offboarding: owners are teams wherever possible, so a departure never leaves a production identity unaccountable.
Risk in Trustivan is explainable — privilege, reach, credential age and exposure, ownership and lifecycle, each a named contribution — never an opaque number. Every finding states its reason and carries the source records behind it.
Every finding explained: the problem, why it matters here, the factors, the recommended action and the evidence.
Source records attached, with timestamps: provider snapshots, sync runs and ingested activity.
Remediation from the finding: assign an owner or quarantine in the platform, and propose provider changes for a second person’s approval.
Scanners find strings and IAM lists accounts. Neither tells you who owns the thing that authenticates, what it can reach or what it does.
The model
One relationship model behind every finding
Trustivan models a human who owns an identity, which has credentials, which authenticate to services, which can access resources, on which the identity performs actions that are allowed or denied. Agents extend the chain with the tools an operator entitled them to use; no agent-to-agent edge exists.
Identity: has credential, owned by, authenticates to, can access, performs action on.
One identity, many credentials — and one credential shared by many identities. Both are findings, and both are visible only on the graph.
Every finding is a path, which is why a credential finding knows its identity and an identity finding knows its owner.
Should this identity be allowed to perform this action against this resource, right now, with this authority?
Identity
a principal that can be authenticated and authorized — a service account, workload, OAuth app, CI job or agent
Credential
what an identity presents to prove it is that identity — a key, token, certificate or key pair
Owner
the human or team accountable for the identity existing and behaving
How Trustivan types each identity, records who owns it and on whose word, and tracks where it is in its life — with evidence behind every value and unknown shown as unknown.
Four weeks, four outcomes: a complete inventory, a classification, an accountable owner for every production identity and an evidence trail — without boiling the ocean.