Resources
Central library of IntentGate resources. The category page is the conceptual anchor; the six controls pages are the technical depth; the standards page is the audit-and-procurement reference; the product pages are what shipped.
Category and concept
Definitions and the broad picture for the agent runtime authorization control category.
What is Agent Runtime Authorization?
The canonical definition of the category — what it is, the threat model, the six required controls, how it fits with your existing security stack. Start here.
Glossary
60+ definitions covering the agent runtime authorization vocabulary: capability tokens, intent enforcement, destructive verbs, memory provenance, MCP, Rego, and related concepts.
The Six Controls
Deep-dive pages on each of the six controls a complete agent runtime authorization solution must implement.
Capability Tokens
HMAC-bound credentials that scope an agent to specific tools, tenants, and time windows. The foundational control.
Intent Enforcement
The gateway control that matches the agent's resolved tool call against the user's declared structured intent. The primary defence against prompt injection.
Policy Engine
Rego-based policy for destructive-verb deny-lists, bulk-row ceilings, value thresholds, and approved-destination lists.
Budget Tracking
Per-tenant token and cost ceilings, fail-closed. The defence against unbounded consumption.
Memory Provenance
Signed chain-of-custody on agent memory writes so high-stakes reads can require verified provenance. Defeats memory poisoning.
Bidirectional PII Filtering
Content inspection on prompts and responses. Eighteen built-in PII classes, counts-only audit so matched values are never persisted.
Standards and threat models
How agent runtime authorization maps to the frameworks security teams, auditors, and procurement reference during evaluation.
Standards Alignment
Per-standard mapping for OWASP Top 10 for LLM, OWASP Top 10 for Agentic AI, NIST AI RMF, MITRE ATLAS, EU AI Act, ISO/IEC 42001, SOC 2, ISO 27001, GDPR.
OWASP Coverage
The CISO-readable OWASP coverage matrix. 14 risks directly mitigated, 1 partial, 5 explicitly out of scope by deployment model.
Product
What IntentGate is and how it works in practice.
Why IntentGate
The four-control bypass essay. The sharpest single positioning page on the site.
Product
What ships, what it does, where it fits.
See it work
Live evaluation lab — a representative agent deployment with all controls enabled and audit chain visible end-to-end.
How it works
Five-check authorization pipeline walkthrough with the install demo cast.
Security & Compliance
Tamper-evident audit, RBAC, JIT elevation, evidence for SOC 2, ISO 27001, GDPR, AI Act.
Open Source
Apache 2.0 licensed gateway, contribution model, the relationship between the open-source core and IntentGate Pro.
Documentation
Operator-facing reference for deploying and operating IntentGate.
Customer-facing PDFs
The following PDFs are available on request through contact. They are the deeper deliverables that complement the public pages: the OWASP-anchored vendor security pack, the CISO-facing threat stories doc, the operations overview, and the architectural comparison battle cards.
- IntentGate Product Brief (v1.9) — the customer-facing introduction in PDF form.
- IntentGate Threat Stories — the CISO walk-through organised around the CIA triad, with the OWASP-anchored coverage map.
- IntentGate Vendor Security Pack (v1.1) — deep technical pack with full OWASP coverage matrix.
- IntentGate Operations Overview — CISO-facing operational commitment doc.
- IntentGate vs. Zscaler & Microsoft Defender — architectural comparison for prospects evaluating IntentGate alongside their perimeter stack.
- IntentGate vs. Azure API Management — architectural comparison for prospects evaluating IntentGate alongside their existing API gateway.