vCISO Lite Launches APRI™ — the MCP Surface for Risk Intelligence, With 82 Tools (Read and Write) and Claim-Grade verify-* Results No Other GRC MCP Returns

ATLANTA — May 19, 2026 — vCISO Lite today launched APRI™(AI-Powered Risk Intelligence), the Model Context Protocol (MCP) surface for the vCISO Lite platform. APRI exposes roughly 82 tools — read andwrite — at mcp.vcisolite.com with OAuth 2.1, PKCE, and Dynamic Client Registration, and includes a 14-tool verify-* family that returns provenance, freshness, and explicit unverifiable_reasons alongside every claim. The same tool graph sits beneath the in-app APRI side-panel inside vCISO Lite Enterprise. APRI is the substrate Trustworthy Autonomy™ runs on.

The Model Context Protocol shipped in late 2024 as Anthropic’s open standard for connecting AI apps to tools and data, and became the de-facto industry protocol through 2025 — every major AI IDE, agent framework, and enterprise-AI vendor added MCP support. The safety story landed alongside. The OWASP Top 10 for Agentic AI Security (December 9, 2025) named tool misuse (ASI02) and identity/privilege abuse (ASI03) as top production risks — both direct MCP concerns. The Databricks AI Security Framework (DASF) v3.0 (March 2026) added agentic AI as its 13th canonical system component and specifically identified MCP as an attack surface with new risks around tool misuse, credential exposure, and privilege escalation. In April 2026, Vanta became the first mainstream GRC vendor to ship an MCP — a 14-tool, read-only compliance catalog. Competent, and narrow.

“Customers are already using their AI clients to ask compliance, vendor, and risk questions — and getting confidently-wrong answers back,” said Yolonda Smith, founder of vCISO Lite. “APRI’s answer isn’t to force the AI to cite. It’s to give the AI the material it needs to answer correctly in the first place. Every verify-*response carries the evidence, the freshness, and an explicit list of anything the platform couldn’t verify — the exact envelope an auditor already wants and SEC cyber disclosure will require. What the AI client does with that envelope is the client’s discipline. When Trustworthy Autonomy is in the loop, honoring it is enforced at the agent layer by cryptographic signature. APRI is the substrate; TA is the enforcement.”

What’s in the release

APRI enters general availability today across two consumption modes over the same underlying tool graph. The release includes:

• Roughly 82 MCP tools, read and write, at mcp.vcisolite.com.Approximately 68 core tools span compliance controls, findings, vendor risk, scanner, red-team, policies, reporting, evidence, work management, audit query, maturity, risk-mapping, business-context, vendor-incident, notifications, corrections, and organization-switching. Nineteen are mutating — attest_control, create_remediation, triage_finding, accept_risk, assign_finding, attach_evidence_to_control, submit_vendor_assessment_for_review, and more. This is not a read-only catalog.
• The 14-tool verify-*family — claim-grade primitives no other GRC MCP returns. Instead of dumping raw data and asking the AI to play forensic accountant, each verify-*tool takes a structured claim (“MFA enforced for production access,” “evidence for SOC 2 CC6.1 collected in the last 90 days,” “vendor risk management active”) and returns a verified result with the evidence, the freshness timestamp, and — where the platform cannot confirm the claim — an explicit unverifiable_reasons list. This is the RFC-042 verified- claim pattern, and it is the ground truth for how an APRI answer differs from a scraped-dashboard answer.
• The in-app APRI side-panel inside vCISO Lite Enterprise. The conversational agent sits inside the vCISO Lite web shell and calls the same MCP catalog beneath. Included in the Enterprise tier with a $25/day per-user Anthropic-spend cap enforced at the gateway.
• OAuth 2.1 with PKCE and Dynamic Client Registration. Two client patterns are live today. Engineering teams on Business tier and above create OAuth clients from their organization’s settings and call the MCP catalog from scripts, CI/CD, IDE extensions, and custom agents. Consumers of AI marketplaces — Claude, Cursor, Claude Desktop, Claude Code — sign in via Zitadel through the authorization-code flow, consent to an organization, and APRI’s tools become callable in-app. Both paths inherit the customer’s vCISO Lite plan-tier gating.
• Layered authorization enforced by construction. Every tool call passes through Layer 1 (plan-tier entitlement), Layer 2 (RBAC), and Layer 3 (per-resource OpenFGA). The multi-tenant boundary is a first-class enforcement point, not a policy: the JWT the gateway accepts is scoped to a single organization, agents cannot escalate their own privilege by asking, and tenant-switch tools (list_my_organizations, get_current_organization, switch_organization) make the current context inspectable at every turn.
• The substrate under Trustworthy Autonomy. APRI itself is not autonomous — it exposes a tool graph and returns claim-grade results. What the AI client does with those results is the client’s discipline. Trustworthy Autonomy™ runs on top of APRI: the agent proposes an action, the action is cryptographically signed (Ed25519), bound to a specific control, backed by a reasoning trace, and re-verified against fresh evidence. That enforcement lives at the agent layer, not at APRI.

Risk Intelligence MCP, not compliance MCP

Vanta’s April 2026 MCP launch made one thing structurally clear: compliance is not the ceiling of what an MCP for security and risk can do — it is one column in a wider surface. vCISO Lite is not racing Vanta on compliance operations. APRI is being built out against a set of risk-intelligence categories most compliance MCPs cannot reach:

• FAIR-style quantification— loss-event frequency and loss magnitude decomposed the way boards, deal teams, and cyber insurers already speak.
• KRIE — Continuous Indicators. Key risk indicators as live signals, not quarterly slides.
• ORE — the Operational Resilience Engine (Keystone). Business function → process → system → people → vendor → facility dependencies as a scenario-testable graph, aligned to DORA, NYDFS Part 500, NIST CSF 2.0, and ISO 22301.
• DC-TPIR — Dependency-Centric Third-Party Incident Response (Refraction).Vendor incidents evaluated against the customer’s own dependency graph — not a generic feed.
• QCD — Quantitative Cyber Diligence.The M&A cyber-risk framework published in the founder’s 2026 book Someone Else’s Debt, addressable from any MCP client.
• Verify-* universalization.The RFC-042 pattern extended across every service APRI queries — so every claim, in every category, carries the same provenance-and-freshness envelope by construction.

Built to the emerging agentic-AI safety standards

APRI’s engineering was designed against the standards that codify what production-grade AI in a privileged environment actually requires:

• OWASP Top 10 for Agentic AI Security— published December 9, 2025 by 100+ researchers and reviewed by a board including NIST, the European Commission, and the Alan Turing Institute. APRI mitigates ASI02 (tool misuse) and ASI03 (identity/privilege abuse) structurally: tools are exposed as a fixed schema with authorization checks per call, and every JWT the gateway accepts is scoped to a specific organization tenant.
• Databricks AI Security Framework (DASF) v3.0— released March 2026 with agentic AI added as the framework’s 13th canonical system component. APRI’s answer to the MCP-specific attack surface DASF names is the same discipline the rest of the vCISO Lite platform enforces: layered authorization, audited tool calls emitted to a signed audit chain, and a multi-tenant boundary implemented as first-class enforcement.
• ISO/IEC 42001— the international AI management standard. Requires an AI management system with lifecycle records, a distinct AI system impact assessment in addition to the AI risk assessment, and the plumbing to inform affected parties when an AI decision touches them. First mid-market audits land in 2027; vCISO Lite’s analysis of what those audits will require is in the company blog’s AI Governance series, starting with “The First AI Audits Hit in 2027. Most Mid-Market Companies Will Fail Them.”
• SEC cyber disclosure 2.0. The provenance, freshness, and unverifiable_reasons envelope the verify-*family returns is what a disclosure regime moving beyond “materiality declared” will require next. APRI is engineered for that horizon, not just today’s standards.

The deeper analysis of how grounding material versus grounding discipline separates good compliance AI from confidently-wrong compliance AI is in the accompanying blog piece, “How often is your compliance AI actually right?” The audit-side counterpart, covering how AI logging requirements will land on assurance firms and their clients, is in “EU AI Act Article 12: What AI Logging Requirements Mean for Audit Firms and Their Clients” and “The 2027-2030 AI Audit Market: How Assurance Firms Will Differentiate.”

Availability

APRI is generally available today. There is no standalone APRI subscription — consumption is gated by vCISO Lite plan tier. Three access modes are live:

• In-app APRI side-panel— included in vCISO Lite Enterprise. $25/day per-user Anthropic-spend cap enforced at the gateway.
• OAuth client_credentials for engineering teams — available on Business tier and above. Client credentials are managed in organization settings, with per-client rate limits and abuse detection at the gateway.
• OAuth authorization_code with PKCE for AI marketplaces— Claude, Cursor, Claude Desktop, Claude Code, and any client that speaks MCP. Users sign in through Zitadel, consent to an organization, and APRI’s tools become callable in-app under the customer’s inherited plan-tier gating.

Install snippets, the developer reference, and the demo video are at vcisolite.com/titanium. Release notes are at the changelog. The wider set of briefs and academic papers is at vcisolite.com/briefs-and-specs.

About vCISO Lite

vCISO Lite is a compliance and cyber risk platform for growing companies that don’t have a full-time CISO. The platform helps customers close compliance gaps across every framework mapped in the Secure Controls Framework(SOC 2, ISO 27001, PCI DSS, HIPAA, DORA, NYDFS Part 500, GDPR, NIST 800-53 and 800-171, FedRAMP, CMMC, and more), quantify cyber risk in the language their board and deal teams already speak, and — with APRI — make the entire platform addressable from the AI clients their teams already use, with claim-grade verify-* results (evidence, freshness, and explicit unverifiable_reasons) returned alongside every call. vCISO Lite is headquartered in Atlanta, Georgia. Learn more at vcisolite.com, read related product releases in the changelog, or explore the AI Governance editorial series.

Media Contact

Press & Analyst Inquiries
Yolonda Smith, Founder
press@vcisolite.com

###