Security

Intutic is a governance layer that sits between your AI agents and LLM providers. Security is foundational — every design decision prioritizes the principle that governance infrastructure must never become the attack surface it's meant to protect.

Threat Model

Intutic defends against three categories of threat:

Threat	Vector	Mitigation
Agent misuse	AI agent executes destructive tool calls (file deletion, credential exposure, unauthorized API calls)	Real-time SOP evaluation with BYPASS/ENHANCE/HIJACK/KILL verdicts via the circuit breaker
Credential leakage	API keys, tokens, or secrets exfiltrated through agent output or prompt injection	DLP scanning in the proxy hot path; secrets detected and redacted before reaching the LLM
Unauthorized access	Rogue agents or developers bypassing governance controls	RBAC enforcement, OBO token scoping, harness config drift detection with auto-revert, and break-glass overrides

Data Flow

Developer → AI Agent → Intutic Proxy (:4000) → LLM Provider
                            │
                      ┌─────▼──────┐
                      │  Policy    │
                      │  Engine    │
                      │  (WASM +   │
                      │   SOPs)    │
                      └─────┬──────┘
                            │
                      ┌─────▼──────┐
                      │  Control   │
                      │  Plane     │
                      │  (:3001)   │
                      └────────────┘

Every LLM request from an AI agent is routed through the local Intutic proxy on port 4000
The proxy evaluates tool calls against SOPs in the WASM policy engine (< 5ms typical)
Verdicts, traces, and telemetry are sent to the control plane for storage and analysis
The proxy never stores prompts or completions — only structured telemetry (tool names, verdicts, token counts, timing)

Data Locality

The proxy runs locally on the developer's machine. LLM traffic is not rerouted through Intutic's cloud infrastructure — it flows directly from the proxy to the LLM provider. Only governance telemetry reaches the control plane.

Encryption

In Transit

All network communication is encrypted with TLS 1.2+:

Path	Protocol
Proxy → LLM Provider	HTTPS (TLS 1.2+)
Proxy → Control Plane	HTTPS (TLS 1.2+)
Sync Daemon ↔ Control Plane	WSS (WebSocket over TLS)
Dashboard → Control Plane	HTTPS (TLS 1.2+)

At Rest

Data	Storage	Encryption
Governance traces	PostgreSQL	AES-256 (managed encryption)
Session state & caches	Valkey (Redis-compatible)	Ephemeral — not persisted to disk
SOP definitions	PostgreSQL	AES-256 (managed encryption)
API keys & credentials	PostgreSQL + system keychain	AES-256 + OS-level keychain encryption
Local config	`~/.intutic/credentials.json`	File permissions (`0600`) + system keychain

Authentication & Identity Enterprise

Intutic provides enterprise-grade identity and access management:

Single Sign-On (SSO) — OIDC integration with Okta, Microsoft Entra ID, Google Workspace, Ping Identity, and any custom OIDC provider
API Keys — Virtual keys (vk_ prefix) for programmatic access, with rotation and revocation
OBO Tokens — Short-lived, employee-scoped agent credentials that replace shared API keys
RBAC Roles — Five-tier role hierarchy (Owner → Admin → EM → Developer → Viewer) with feature-level gating

→ Full details: Security & Identity guide

Compliance Enterprise

Intutic helps your organization meet regulatory requirements for AI governance:

SOC 2 Readiness — Automated evidence collection for trust service criteria (security, availability, processing integrity, confidentiality)
HIPAA BAA — PHI safeguard tracking and BAA status management
GDPR — Data erasure (right to be forgotten), consent management, and data processing agreements
Data Residency — Pin governance data to US, EU, or APAC regions

Enforcement Coverage

Connected harnesses are scored on a four-tier system:

Tier	Enforcement Level
A	Strict — immutable rules, credential redaction, system command blocking
B	Moderate — prompts for suspicious tool combinations
C	Audit — records traces without blocking
D	Observe — alerts only, fail-open

→ Full details: Security & Identity guide

Infrastructure Security

Layer	Controls
Network	VPC isolation, private subnets, no public database endpoints
Compute	GKE with node auto-upgrade, workload identity, pod security standards
Secrets	GCP Secret Manager — never committed to source control
Monitoring	Structured logging with trace context
Supply chain	Dependabot, lockfile integrity, signed container images

→ Self-hosted deployment: Deployment guide → Stream security events: Stream Alerts guide

Responsible Disclosure

If you discover a security vulnerability, please report it to security@intutic.ai. We aim to acknowledge reports within 24 hours and provide a fix or mitigation within 7 business days.

Page	What it covers
Security & Identity	SSO setup, API keys, OBO tokens, RBAC roles, member management
Emergency Overrides	Break-glass override tokens, request workflows, audit logs
How It Works	Full architecture walkthrough — proxy, policy engine, control plane
Stream Alerts	Stream governance events to Slack channels or custom webhook endpoints
Settings	Dashboard settings — security tab, compliance tab, API key management

Security ​

Threat Model ​

Data Flow ​

Encryption ​

In Transit ​

At Rest ​

Authentication & Identity Enterprise ​

Compliance Enterprise ​

Enforcement Coverage ​

Infrastructure Security ​

Responsible Disclosure ​