Skip to main content
Grantex provides technical controls that map directly to the three major regulatory frameworks governing AI agent security. This page documents each requirement and the corresponding Grantex feature.

​
OWASP Agentic Security Top 10

Published December 2025 — the first industry-standard threat taxonomy for autonomous AI agents.
RiskDescriptionGrantex ControlImplementation
ASI-01Agent goal hijackingScoped permissionsJWT scp claim limits actions regardless of agent intent. A hijacked agent cannot exceed its granted scopes.
ASI-03Identity & privilege abusePer-agent cryptographic identityEach agent gets a unique DID. The agt claim binds tokens to specific agents. Tokens are non-transferable.
ASI-05Privilege escalationDelegation invariantsChild scopes must be a strict subset of parent scopes. delegationDepth claim enforced at protocol level.
ASI-10Rogue agents (no revocation)Instant cascading revocationPOST /v1/tokens/revoke blocklists the JTI. All child delegations invalidated in the same operation.

​
EU AI Act

Binding August 2026 — the world’s first comprehensive AI regulation.
ArticleRequirementGrantex ControlImplementation
Art. 9Risk management per AI systemBudget controls + anomaly detectionPOST /v1/budget/allocate sets spending limits. Anomaly detection flags unusual patterns. Policy-as-code (OPA, Cedar) for fine-grained rules.
Art. 13Transparency for autonomous actionsConsent flow + Verifiable CredentialsHuman sees plain-English consent screen. VC-JWTs provide portable proof of authorization. SD-JWT enables selective disclosure.
Art. 14Human oversight — logged & auditablePrincipal dashboard + event streaming + revocationGET /v1/events/stream (SSE) surfaces actions in real time. Principal Sessions dashboard. One-click cascade revocation.

​
NIST AI Risk Management Framework

Active now — required for US government agencies and federal contractors per Executive Order 14110.
ControlRequirementGrantex ControlImplementation
Govern 1.1AI roles & accountabilityGrant-to-human traceabilityEvery JWT carries sub (human principal), dev (developer), and agt (agent). The accountability chain is unbroken.
Map 5.1Agent action attributionAudit trail linked to grantsPOST /v1/audit/log records actions. Each entry references the grantId and agentId. Entries are hash-chained.
Measure 2.5Audit trails for autonomous operationsHash-chained, append-only auditAudit entries are append-only, hash-chained (tamper-evident), filterable by agent/grant/principal/time, and exportable via compliance evidence packs.

​
Full Compliance Matrix

RequirementFrameworkGrantex FeatureVerification Method
Scoped permissionsOWASP ASI-01JWT scp claimService verifies scope before executing action
Per-agent identityOWASP ASI-03DID per agent, agt claimSignature verification against agent’s public key
Delegation depth limitsOWASP ASI-05delegationDepth claimProtocol rejects invariant-violating delegations
Instant revocationOWASP ASI-10Redis blocklist + StatusList2021POST /v1/tokens/verify returns valid: false
Risk managementEU Art. 9Budgets, anomaly detection, policiesAPI endpoints + dashboard monitoring
TransparencyEU Art. 13Consent flow, VCs, SD-JWTConsent URL, credential verification
Human oversightEU Art. 14Events, principal sessions, revocationSSE stream, dashboard, API
AccountabilityNIST Govern 1.1sub + dev + agt claimsJWT decode traces action to human
Action attributionNIST Map 5.1Hash-chained audit entriesGET /v1/audit/entries with filters
Audit trailsNIST Measure 2.5Append-only, exportable auditEvidence pack export via API

​
Compliance Evidence Pack

Grantex can generate a compliance evidence pack that bundles all relevant data for auditors: 
# Generate a compliance evidence pack via CLI
grantex compliance export --format json --from 2026-01-01 --to 2026-03-28

# Or via API
curl -X POST https://api.grantex.dev/v1/compliance/evidence \
  -H "Authorization: Bearer $API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"from": "2026-01-01", "to": "2026-03-28"}'
The evidence pack includes:
  • All grants issued in the period
  • Token issuance and revocation events
  • Audit trail entries (hash-chained)
  • Agent registrations and delegation chains
  • Policy snapshots
  • Anomaly detection findings

​
Certifications