FAQ · Eight readers, one record

AI agent governance: questions by role.

CISO, CIO and CTO, board and audit, legal, regulators, underwriters, analysts. The fourth A is the same frame through every door. Each asks different questions. One record answers all of them.

/001 · CIO / CTOBEHAVRY · RECORD

What your CTO needs to know.

The CTO owns the deployment. These are the questions that determine whether AI ships to production or stays in a sandbox.

Does this slow down my developers?

No. OPA policy evaluation itself runs in microseconds; total added latency is typically 5–30ms per tool call, most of it the network hop. For context, LLM inference takes 500ms–5s. Developers only see the authorization layer when an action actually violates policy. Safe actions auto-approve at full speed.

How long does deployment take?

Agents point their MCP config at the Behavry authorizer URL. That's a config file change, not a code change. A team can be authorized in an afternoon. Full enterprise rollout, with IdP connectors, SIEM integration, and custom policies, typically takes days, not weeks.

What agents and tools do you support?

12 MCP clients ship a one-click enrollment template (Claude Code, Claude Desktop, Cursor, VS Code, Windsurf, Warp, Zed, Antigravity, Kiro, Codex CLI, OpenAI, Open Interpreter). Any other MCP client speaking Streamable HTTP, SSE or stdio — JetBrains IDEs, Cline and Continue among them — is governed by the same proxy without one. 6 API proxies (OpenAI, Anthropic, Gemini, Ollama, NemoClaw, OpenShell). 12 browser services. 7 vibe-coding platforms. 12 local model runtimes.

Can this run in our infrastructure?

Four deployment models: Full SaaS (fastest start), Hybrid (control plane SaaS, data plane in your VPC), BYOC (Helm + Terraform in your cloud), Self-Hosted (air-gapped). All four share a single data plane image with identical capabilities. Agent traffic stays in your network on Hybrid, BYOC, and Self-Hosted.

Do my developers have to change their code?

Zero agent code changes. Agents point at the Behavry authorizer URL instead of the target MCP server URL. That's a JSON config change. The authorizer is transparent: agents see standard MCP protocol responses. No SDK required, no wrapper functions, no instrumentation.

What does this give me that network-layer controls can't?

Network security tools and gateways vary: some enforce identity-aware, application-level, or per-tool policies. When evaluating your existing stack, check whether it evaluates an individual agent’s requested action before execution, including the agent identity, target resource, and session context. On supported, configured paths, Behavry evaluates each tool call against policy before forwarding it, with allow, deny, or human-approval outcomes and a signed Decision Trace.

/002 · CISOBEHAVRY · RECORD

What your CISO will validate.

The CISO doesn't buy Behavry. But the CISO validates that the authorization layer meets security requirements. These are the questions they'll bring.

How do you handle DLP for AI agent interactions?

33 DLP patterns scan every tool call payload. Patterns include AWS/GCP/Azure credentials, API tokens, private keys, SSNs (with Luhn validation), credit cards, and PII. Critical-severity patterns auto-deny before OPA policy evaluation. The full 4-stage data protection pipeline includes classification, redaction with pseudonymization, BYOK encryption (AES-256-GCM + AWS KMS), and retention purge with immutable decryption audit trail.

What does the audit trail actually look like?

The Decision Trace is not a log. It's a causal chain-of-custody artifact. Every event records the agent identity, the tool call, the authorization decision (allow / deny / intercept), the DLP findings, the behavioral risk score, the workflow session, the parent event, and the causal depth. Events are SHA-256 hash-chained. The chain is verified nightly.

On supported, configured paths, Behavry records the policy decision at the action boundary before the call is forwarded or blocked. The signed, hash-chained Decision Trace connects that decision to the agent identity, requested action, and available context. Verification checks integrity and authenticity under the supplied trust anchor; it does not establish completeness outside the governed path or that the policy was adequate.

How do you detect prompt injection and inbound attacks?

Behavry scans MCP tool call responses before they reach agent context. 16 detection patterns across 7 attack classes: imperative commands, authority claims, permission expansion, role reassignment, encoded payloads, exfiltration instructions, and conditioning sequences. Critical findings trigger Intercept escalation with allow-sanitized, allow-original, or deny options.

What compliance frameworks do you map to?

Thirteen frameworks, 170 controls. SOC 2 (CC6.1–CC6.6, CC7.2, CC7.3, CC8.1, CC9.2), ISO/IEC 27001:2022 (A.5.15, A.5.28, A.8.2, A.8.10, A.8.12, A.8.15, A.8.16, A.8.24), EU AI Act, NIST AI RMF 1.0, HIPAA, PCI DSS v4.0, HITRUST AI, CIS AI/ML, OWASP ASI, AIUC-1, financial services (SR 11-7, NYDFS Part 500, SEC), insurance (NAIC, Colorado, IDFPR), and privacy (GDPR, CCPA, LGPD, PIPEDA). Mapped means our shipped controls satisfy the named clauses — not that an auditor has certified them.

Can agents bypass the authorizer?

In a properly configured deployment, MCP target servers are only reachable through the Behavry authorizer via network segmentation. Direct connections are monitored and flagged. This is standard zero-trust architecture.

How do you handle multi-agent delegation chains?

Behavry's workflow authorization tracks delegation chains with cryptographic session tokens (wf_token JWT). Every tool call in a multi-agent pipeline records its parent event ID, causal depth, and the full delegation chain. Cross-session trust reset detection prevents persistent compromised behavior across session boundaries. Blast radius limits cap the total impact of any single delegation chain.

/003 · BOARD / AUDIT COMMITTEEBEHAVRY · RECORD

How the board will frame it.

The board doesn't ask about OPA policies. They ask about risk, accountability, and whether the organization can prove what happened.

How many AI agents are running in our environment, and who authorized them?

Behavry's AI Surface Discovery answers this directly. 56 AI-capable SaaS platforms are fingerprinted and tracked through a four-state model: Licensed, Enabled, Active, and Governed. IdP connectors (Okta, Azure AD, Google) cross-reference who has access. SaaS admin API connectors confirm what's enabled. The exposure score tells you the gap between what's running and what's authorized.

If an AI agent causes a data breach, can we prove what happened?

Yes, for actions that passed through Behavry. The Decision Trace provides a tamper-evident, hash-chained record of each governed action, including the policy decision, the data classification, and the delegation chain if the action was part of a multi-agent workflow. This artifact is exportable for legal and audit review.

What happens when an agent does something it shouldn't?

It depends on what the policy says. Behavry supports three outcomes: Allow (action proceeds, audit logged), Deny (action blocked, agent receives error, audit logged), and Intercept (action held, human reviewer approves or rejects, agent receives the decision). The global kill switch suspends all agent activity instantly when needed. Per-agent rate limiting tightens automatically as risk scores increase.

Are we compliant with the EU AI Act and other emerging regulations?

Behavry maps directly to EU AI Act Articles 9 (risk management), 13 (transparency), and 14 (human oversight). The Decision Trace provides the transparency artifact. Intercept provides the human oversight mechanism. Behavioral risk scoring provides the risk management framework. Compliance PDF export produces a report formatted for auditors, with framework mapping, control status, and audit statistics. Mapping is not certification, and acceptance is your auditor’s call.

What's our liability if employees are building apps with AI tools we don't authorize?

This is the citizen-coder problem. Employees using Replit, Lovable, Bolt, and similar platforms are shipping production apps without security review. Behavry discovers these apps via browser extension fingerprinting and platform API connectors, scores them across 7 risk signals, and enforces OPA policies, from auto-enrollment to denial. Discovery depends on the sources you connect: apps built from managed devices with the extension, and platforms with a connected API. An app that is discovered and still ungoverned after 30 days raises an alert for escalation.

How do we know this authorization layer itself is trustworthy?

The Attestation Separation Principle: any entity that can act cannot independently attest to its own behavior. Behavry is architecturally independent from the agents it authorizes. It doesn't modify agent code, doesn't run inside agent processes, and doesn't depend on agent self-reporting. The authorizer sits in the network path. The audit chain is hash-verified. The authorization layer is provably separate from the thing being authorized.

CISO gets accountability/Board gets oversight/Team gets to ship

/004 · OPERATORS AND PARTNERSAUTHORITY · COVERAGE · DELIVERY

Authority, coverage, and who runs it.

The questions operators and providers ask once agents are doing real work.

Who approves a change in an agent's authority?

A named operator with authority the customer has delegated. Behavry can recommend a change from the evidence; it does not approve its own recommendation. Recommendation, approval, and the applied grant are separate, recorded states. Earned Autonomy →

What does shadow mode change?

Nothing about permissions. In the planned A1 to A2 assessment, Behavry records a recommendation while the agent keeps working at its current level. Automatic promotion in production is not offered.

Does a higher autonomy level override other restrictions?

No. A level binds a restriction profile; OPA policy, DLP, zones, identity, incident restrictions, and tenant controls still apply, and the tighter rule wins. A3 means policy alone, not no policy.

What does Behavry not see?

Actions that do not pass through a supported, configured execution path. Coverage depends on which integrations you configure, and on network controls that keep agents from going around them. Coverage →

Can our MSP or MSSP run this for us?

Yes. A provider can operate Behavry across your environment within the authority you delegate, with approval rights and evidence access staying with you. Providers can start with AI observation and ongoing monitoring, and add enforcement only with your approval, or begin with a scoped 30-day enforcement pilot. For MSPs and MSSPs → · Observation without blocking: a guide →

What does verifying a Decision Trace actually check?

That the evidence package is intact and was signed by the key in the trust anchor you supply. It does not show that events outside the governed path did not happen, or that a policy was adequate. Verify a sample →

/005 · EVALUATING BEHAVRYWHAT TO VERIFY

What to verify when you evaluate Behavry.

Neutral criteria to test in a scoped evaluation, rather than take on trust. Ask for each against your own environment.

Integration coverage

Which of your agent surfaces, clients, and target systems are on a supported, configured path today, versus proposed or subject to a compatibility assessment. Protocol compatibility is not the same as a tested workflow. Integrations →

Action interception

That Behavry evaluates the action before it runs on the paths you configure, not just records it after. Test an action that should be denied and confirm it does not reach the target system.

Policy behavior

How allow, escalate, and deny decisions are expressed, and that the same context produces the same decision. Behavry evaluates on a deterministic OPA engine with no LLM in the decision path.

Approvals

That a held action waits for a named human, and that a recommendation is never a self-approval. Recommendation, approval, and the applied grant are separate, recorded states.

Evidence access

That you can export a Decision Trace and verify it with a trust anchor you supply, with no Behavry account, and that access survives a change of provider. Verify a sample →

Deployment requirements

What a connected path requires in your environment, which deployment model applies, and where the current product is versus its roadmap. Transparency →

Operational responsibilities

Who runs the platform, who reviews alerts, and who holds approval authority, especially when a provider delivers the service. The operating model →

/ONWARDBEHAVRY · RECORD
  • THE DECISION TRACE

    What the agent did, why it was allowed, and a record anyone can verify without us.

  • ARCHITECTURE

    One vantage point in the path of the action, across supported agent surfaces.

  • COVERAGE

    Eight ways an agent gets in, and where each one is intercepted.

The authorization layer that lets everyone say yes.

BOOK A DEMO →MSP or MSSP? Discuss an engagement →