Skip to main contentSkip to footer

A secure architecture for the Claude API in the enterprise

A reference architecture for the Claude API in regulated firms: one gateway, no static keys, private routes, approval gates and logs an auditor can follow.

YatiSphere team15 min readPublished November 17, 2026Facts checked November 17, 2026

You can get Claude answering questions in an afternoon. Getting it through a security review at a bank or a law firm is a different job. The reviewers ask the same five things every time: who can call the model, which company processes the data, what gets kept and for how long, what an agent is allowed to change, and how you would find out if something went wrong.

Below is the architecture we use to answer them. It holds up whether you reach Claude through Anthropic’s own API, Claude Platform on AWS, Amazon Bedrock, Google Cloud or Microsoft Foundry, and it leaves you free to change route later. If you haven’t picked a route yet, start with our comparison of Claude enterprise deployment options.

A quick warning: the provider details here are correct as of November 2026, and they move fast. Check anything you depend on against the current docs before you build.

1. Give Claude one front door

If you only take one idea from this piece, take this one: no application should call a model provider directly. Every call goes through an AI gateway that you own and run inside your network. The gateway works out who is calling, applies your data and spend rules, writes a log line, and sends the request down the one route that caller is cleared to use.

Agents need a second door, for tools. That is the MCP gateway, which checks the agent’s identity and its tool list before anything touches a system of record. Logging, evaluation, cost control and incident runbooks don’t belong to any one box in the drawing. They watch all of them.

  • You probably own a gateway already. Azure API Management, Amazon API Gateway, Apigee and Kong all do the job, and so do dedicated LLM gateways. Spend your time on the policy; the product choice matters much less.
  • With routing and policy in one place, moving a workload from one Claude route to another becomes a config change. Nobody has to touch application code.
Logical architecture for Claude in a regulated enterpriseA layered view. Clients (event sources, employees and developers) call into a VPC. In the app subnet, which has no internet route, an agent runtime and an app service run. In the gateway subnet, an MCP gateway handles tool calls and an AI gateway runs nine checks on every model call: authenticate, authorize, classify, redact PII, screen input, meter quota, route, check output and log with a trace ID. The data subnet holds MCP servers in front of systems of record. The egress subnet holds a private endpoint to the cloud routes and an egress proxy to the Claude API. Below sit the five model routes with their data processor, and the shared services: identity provider, secrets vault, SIEM, evaluation and cost control. A table lists the nine interfaces with protocol and authentication.FIG 1LOGICAL ARCHITECTUREany Claude route · rev 1.0 · Nov 2026ClientsCorporate networkApp subnet10.20.10.0/24No internet routeGateway subnet10.20.20.0/24Your controlsData subnet10.20.30.0/24Egress subnet10.20.40.0/24Model routesExternalShared servicesPlatform teamYOUR VPC · 10.20.0.0/16Events, schedulesQueue, cron, webhooksEmployeesBrowser, Teams, SlackDevelopersClaude Code CLI, IDEAgent runtimeOne identity per agentApp serviceCopilot, RAGMCP gatewayOAuth 2.1 resource serverTool allow-list per agentApproval hook for writesKill switchAI gatewayAPI management or LLM gateway1authenticate2authorize3classify4redact PII5screen input6meter quota7route8check output9log + trace IDMCP serversDMS, search, core systemsPrivate endpointPrivateLink · PSC · Azure PEEgress proxyHost allow-listprivate network pathpublic TLSClaudePlatform on AWSAnthropicAmazonBedrockAWSGoogle CloudGoogleMicrosoftFoundryAnthropicClaude APIAnthropicIdentityEntra ID, OktaSecrets vaultGateway, admin keysSIEMAll logs, trace IDEvaluationTests per changeCost controlQuotas, budgetsINTERFACESEvents → agent runtimeQueue message; workload identityEmployees → app serviceHTTPS; SSO (OIDC) with MFADevelopers → AI gatewayClaude Code; per-user identityApp, agent → AI gatewayHTTPS; OIDC JWT, no API keyAgent → MCP gatewayOAuth 2.1; RFC 8707 audienceMCP gateway → MCP serversToken bound to one serverAI gateway → cloud routesPrivate endpoint; cloud IAMAI gateway → Claude APIEgress proxy; TLS 443; WIF tokenEvery component → SIEMLogs carrying one trace IDCIDRs are examples. Bar on each route: dark navy, Anthropic processes the data; light blue, the cloud provider does.
Logical architecture. Numbers match the interface table: protocol and authentication on every hop.

2. Pick the route by who processes the data

Before you compare features, settle who is allowed to process your data. That question alone usually knocks out half the options. On the Claude API, Claude Platform on AWS and Microsoft Foundry, Anthropic processes prompts and outputs. On Amazon Bedrock and the Google-managed version of Google Cloud’s Gemini Enterprise Agent Platform, the cloud provider does.

Nothing stops you running several routes, one per data class, all behind the same gateway. Whichever route you land on, check its region, retention mode and features before you build.

Figure 2 sets the five routes side by side. To pick one, work through these in order and stop at the first match:

  • Only your cloud provider may process the data: Amazon Bedrock, or the Google-managed version of Google Cloud.
  • You need a HIPAA BAA with Anthropic: the Claude API with a HIPAA-enabled organization.
  • You want AWS IAM and AWS billing: Claude Platform on AWS.
  • You are standardized on Microsoft Azure: Microsoft Foundry.
  • None of the above: the Claude API direct, where new models and features usually land first.
Claude route matrix, November 2026A table of the five Claude routes. Claude API: Anthropic processes, workload identity federation, egress proxy over public TLS, Console usage and activity APIs, HIPAA with a BAA and a HIPAA-enabled organization. Claude Platform on AWS: Anthropic processes, AWS IAM with SigV4, PrivateLink endpoint aws-external-anthropic, CloudTrail data events, no HIPAA. Amazon Bedrock: AWS processes, AWS IAM with SigV4, PrivateLink endpoint bedrock-mantle or bedrock-runtime, CloudTrail, AWS terms apply. Google Cloud, Google-managed: Google processes, Google Cloud IAM service account, Private Service Connect with VPC Service Controls, Cloud Audit Logs, Google terms apply. Microsoft Foundry: Anthropic processes, Entra ID, Foundry resource in an Azure VNet, Azure Monitor, HIPAA with the HIPAA configuration.FIG 2ROUTE MATRIXchecked Nov 17, 2026RouteProcessorWorkload authPrivate pathAudit trailHIPAA via AnthropicClaude APIdirectAnthropicWorkload identityfederation (OIDC)Egress proxy,public TLS 443Console usage, costand activity APIsYes: BAA andHIPAA-enabled orgClaude Platformon AWSAnthropicAWS IAM,SigV4 signingAWS PrivateLinkaws-external-anthropicCloudTraildata events (opt-in)NoAmazon BedrockAWSAWS IAM,SigV4 signingAWS PrivateLinkbedrock-mantle/runtimeCloudTrailNot covered;AWS terms applyGoogle CloudGoogle-managedGoogleGoogle Cloud IAMservice accountPrivate ServiceConnect + VPC-SCCloud Audit LogsNot covered;Google terms applyMicrosoft FoundryAnthropicMicrosoftEntra IDFoundry resourcein an Azure VNetAzure MonitorYes: HIPAAconfigurationAnthropic processes prompts and outputsThe cloud provider processes themResidency: Claude Platform on AWS offers per-request US-only inference at 1.1x; Foundry offers Global and US Data Zone deployments.Anthropic-managed Claude on Google Cloud is a separate option with Anthropic as processor. Confirm region and features before you build.
Route matrix: processor, workload identity, private path, audit trail and HIPAA coverage for each route.

3. Get rid of static API keys

Many AI credential leaks start with a long-lived API key. Someone copies it into a CI variable, pastes it into a notebook, or commits it by accident, and it works for anyone who finds it. The better pattern is to give each workload a short-lived identity from the identity provider you already run, and let the gateway hold its own credential for the provider.

On the Claude API, that means workload identity federation. The workload presents a JWT from AWS IAM, Google Cloud or any standards-compliant OIDC issuer (GitHub Actions, Kubernetes, Microsoft Entra ID and Okta all qualify), and Anthropic swaps it for a short-lived token. You can set the token lifetime anywhere from 60 seconds to 24 hours; the default is one hour. On the cloud routes, a cloud IAM role does the same job, and Claude Platform on AWS uses SigV4 signing as its main method.

Watch for one thing during migration. If an old ANTHROPIC_API_KEY environment variable is still set, it quietly overrides federation, and everything looks fine until someone audits it. Unset it everywhere the workload runs.

  • People sign in through Microsoft Entra ID or Okta with MFA. Console and cloud admin roles use named accounts, never shared ones.
  • Workloads reach the gateway with OIDC tokens from your identity provider.
  • Each agent gets its own identity. Don’t let agents borrow a shared service account or a person’s token; we explain why in every AI agent needs an identity.
  • Create one Claude Console workspace per environment and data class. Each has its own members, keys, rate limits and spend limit, and a key scoped to a workspace can reach only that workspace. Use workspace-scoped keys, or service accounts with explicit workspace memberships.
  • Keep Admin API keys (sk-ant-admin...) in a vault, and let only automation use them.
ANTHROPIC_FEDERATION_RULE_ID=fdrl_...
ANTHROPIC_ORGANIZATION_ID=<organization uuid>
ANTHROPIC_SERVICE_ACCOUNT_ID=svac_...
ANTHROPIC_WORKSPACE_ID=wrkspc_...
ANTHROPIC_IDENTITY_TOKEN_FILE=/var/run/secrets/anthropic.com/token
# and make sure ANTHROPIC_API_KEY is unset
Workload identity federation for the Claude SDK, configured through environment variables. The workload holds no API key.

4. Lock down the network

Model traffic should have exactly two ways out of your network. One is a private endpoint to a cloud AI service. The other is an egress proxy that allows the approved provider API and nothing else. Your applications and agents live in a subnet with no internet route at all, which means the only way they can reach Claude is through the gateway.

  • On AWS, create an interface endpoint for the Bedrock endpoint you call, com.amazonaws.<region>.bedrock-mantle (Messages API) or com.amazonaws.<region>.bedrock-runtime (InvokeModel and Converse), or for com.amazonaws.<region>.aws-external-anthropic (Claude Platform on AWS). Turn on private DNS and the SDK keeps its normal base URL. On Bedrock, an endpoint policy can then limit which principals and models are allowed.
  • On Google Cloud, use a Private Service Connect endpoint for Google APIs and put the project inside a VPC Service Controls perimeter.
  • On Azure, place the Foundry resource in an Azure Virtual Network.
  • For the direct Claude API, run an egress proxy that allows only the provider API host and logs every connection. Block other AI endpoints by default too. It stops people quietly wiring unapproved AI tools into your network.
  • Pin each route to the regions your rules allow. Claude Platform on AWS can run US-only inference per request at 1.1x the price, and Foundry offers Global and US Data Zone deployments.
Network view for Claude traffic, AWS exampleThe corporate network connects to the VPC over Direct Connect or VPN. The VPC has four subnets. The app subnet runs apps and agents with a local-only route table and may only send HTTPS to the gateway security group. The gateway subnet runs the AI and MCP gateways behind an internal load balancer. The egress subnet runs an egress proxy that allows api.anthropic.com, routing out through a NAT gateway and an internet gateway to the Claude API. The endpoint subnet holds interface endpoints with private DNS for bedrock-mantle, bedrock-runtime and aws-external-anthropic, so traffic to Amazon Bedrock and Claude Platform on AWS stays on the AWS network. On Google Cloud the same pattern uses a Private Service Connect endpoint inside a VPC Service Controls perimeter; on Azure, a Foundry resource in a VNet.FIG 3NETWORK VIEWAWS example · rev 1.0Corporate networkUsers, developersDirect Connect or VPNVPC 10.20.0.0/16 · APPROVED REGIONApp subnet10.20.10.0/24rt: local onlysg: out 443 → gwGateway subnet10.20.20.0/24rt: local onlysg: in 443 ← appEgress subnet10.20.40.0/24rt: 0/0 → nat-gwsg: in 3128 ← gwEndpoint subnet10.20.50.0/24rt: local onlysg: in 443 ← gwApps and agentsECS or EKS tasksAI + MCP gatewaysInternal ALB, 2 AZsEgress proxyapi.anthropic.comPublic subnetNAT gatewayInterface endpointsprivate DNS onIGW4433128Same pattern elsewhereGoogle CloudPrivate Service Connectendpoint inside a VPCService Controls perimeter.AzureFoundry resource placedin an Azure VNet.Direct Claude APIOnly via the egress proxy.PUBLIC INTERNETClaude APIapi.anthropic.comAWS NETWORKAmazon Bedrockbedrock-mantle/runtimeClaude Platformaws-external-anthropicYour VPCProvider networkAllowed flow (HTTPS unless marked)Icons: AWS Architecture IconsCIDRs, ports and security group names are examples. Only the egress subnet has a default route; one AZ shown, deploy in two or more.
Network view on AWS: subnets, route tables and security groups. Only the egress subnet has a default route.

5. What the gateway does with each call

The gateway runs the same checks, in the same order, on every request. The important ones are strict: if a check can’t run, the request doesn’t leave. A logging outage should stop traffic rather than let it through unrecorded.

  • Authenticate the caller and tie it to an app, a team and, where relevant, an agent.
  • Confirm that caller may use this model, route and data class.
  • Classify the request as public, internal, confidential or restricted, using the app’s registration plus content checks.
  • Strip or tokenize PII and secrets the use case doesn’t need. This shrinks what leaves your network, but the processor stays the same.
  • Screen the input for prompt injection and policy breaches. On AWS, Bedrock Guardrails can handle this, and on the bedrock-runtime endpoint IAM can insist on a guardrail for every InvokeModel or Converse call. Guardrails are not available on bedrock-mantle.
  • Count tokens and requests against each app’s quota and monthly budget.
  • Send the request to the approved route and model. A fallback route is fine, as long as it is approved for the same data class.
  • Check the output for leaked secrets, PII and unsafe content.
  • Log the caller, app, model, route, tokens, cost, data class, each policy decision and a trace ID.

6. Know what gets stored, and for how long

Anthropic says retained data is never used to train models without your express permission. Retention itself is the trickier part, because it varies by feature. Under zero data retention (ZDR), Anthropic doesn’t keep prompts or responses at rest once the response is returned. You get ZDR per organization through Anthropic’s sales team, and it covers the Messages and Token Counting APIs for eligible features, Claude Code with Console API keys, Claude Platform on AWS, and Anthropic-managed Claude on Google Cloud.

Several features fall outside ZDR, and your design should account for that. Our rule is simple: anything that needs retention gets its own workspace and its own data-class approval, so a ZDR workspace can never call it by mistake. Also note that if Anthropic’s trust and safety systems flag a session, its inputs and outputs may be kept for up to two years, ZDR or HIPAA notwithstanding.

For HIPAA, you need a signed BAA and a HIPAA-enabled organization. That covers eligible Claude API features, and once you enable it for an API organization, you can’t switch it off. Claude in Microsoft Foundry also needs the HIPAA configuration on your Azure subscription, arranged through your Anthropic account team. It doesn’t extend to Bedrock, Google Cloud’s Agent Platform, Claude Platform on AWS or Claude Code. On Bedrock and Google-managed Google Cloud, the cloud provider’s terms apply instead of Anthropic’s ZDR program.

Feature                      Retention
Batch processing             29 days, not ZDR eligible
Files API                    Until deleted or expired
Code execution tool          Container data up to 30 days
Programmatic tool calling    Up to 30 days
Managed Agents (hosted)      Until you delete transcripts
MCP connector                Standard; not ZDR eligible
Agent Skills                 Standard; not ZDR eligible
Fable and Mythos models      Require 30-day retention
Prompt caching               Memory only; ZDR eligible
Features that retain data outside ZDR, per Anthropic’s data retention docs (November 2026).

7. Agents only act through approved tools

Agents are the point where Claude stops suggesting things and starts doing them, so this is where most of the risk lives. Every tool call goes through the MCP gateway. Read-only tools run with a token scoped to a single server. Anything that changes data waits until a named person approves it.

The current MCP specification (revision 2026-07-28) treats a protected MCP server as an OAuth 2.1 resource server, and clients must use RFC 8707 resource indicators so each token is minted for one specific server. Hold every remote MCP server you allow to that standard, including the ones vendors host for you. Our MCP governance guide goes deeper on the gateway itself.

  • Give each agent the tools its job needs and no more. Adding a tool should trigger a security review and a fresh run of the test suite.
  • Payments, filings, client messages, deletions and permission changes all wait for approval. Our kill switch and approval design walks through how.
  • Documents, emails and web pages the agent retrieves are data. They can’t add tools or rewrite the agent’s instructions, however they are worded.
  • Cap steps, tool calls and tokens per run at both gateways. Limits that live only in the agent’s own code are easy to bypass.
  • Make the kill switch a single action that revokes the agent’s identity, blocks its tools and stops new runs. Practice it every quarter, because a switch nobody has used tends to fail on the day.
Sequence of an agent tool call with human approvalA UML sequence diagram. 1: the user sends a task to the agent runtime. 2: the agent calls the AI gateway with its OIDC token. The gateway authenticates, classifies, redacts, screens and meters the call. 3: it forwards the checked prompt to Claude. 4: Claude returns a tool_use block. 5: the gateway returns the reply with a trace ID. 6: the agent calls the tool through the MCP gateway with an OAuth 2.1 token, and the gateway checks the agent identity, allow-list and step caps. If the tool writes data, 7: the MCP gateway asks a named approver, 8: the approver approves, 9: the write executes. If the tool is read-only, 10: it reads with a token scoped to that server. 11: the tool result, or a denial with the reason, returns to the agent. Every message carries one trace ID to the SIEM; the kill switch revokes the agent identity or disables the tool, which makes calls 2 and 6 fail.FIG 4SEQUENCE: AGENT TOOL CALLwrite path with approvalUser«actor»Agent runtime«service»AI gateway«your control»Claude«model»MCP gateway«your control»Approver«named person»Core system«system of record»1 task request2 messages + JWTauthn, classify, redact, screen, quota3 checked prompt4 tool_use5 reply, trace ID6 tools/call · OAuth 2.1 tokenagent ID, allow-list, step capsalt[tool writes data]7 approval request8 approved9 execute write[read-only tool]10 read, scoped token11 tool_result, or denial + reasonKill switch: revoke the agent’s identity or disable the tool at the MCP gateway. Calls 2 and 6 then fail.Every message carries one trace ID to the SIEM, including the approver’s decision.
Sequence for one agent tool call. Reads run with a scoped token; writes wait for a named approver.

8. Leave a trail an auditor can follow

Pick any decision Claude helped with. An auditor should be able to see who asked, what Claude was given, what it said back, which tools it called and who signed off on the result. To get there, carry one trace ID from the user’s request through the gateway, every model call, every tool call and the approval record, and send all of it to your SIEM.

Evaluation belongs in the same picture. Agree acceptance tests and pass marks with the business owner before you build. Rerun them whenever the model, prompt, tools or retrieval data change, and sample production answers every week. Testing AI agents before production sets out the test plan.

  • From the gateway: caller, app, agent, model, route, tokens, cost, data class and policy decisions.
  • Prompt and response text only where the use case needs it, encrypted, and kept for the shortest time you can defend.
  • From the providers: CloudTrail for Bedrock and Claude Platform on AWS (inference on Claude Platform on AWS and on bedrock-mantle is logged as data events you have to switch on; bedrock-runtime inference is logged as management events by default), Google Cloud audit logs, Azure Monitor, and the Claude Console usage, cost and activity APIs.
  • Every admin change to keys, workspaces, members and limits, with alerts on them.

9. Treat Claude Code as privileged tooling

Claude Code runs on developer laptops with access to source code, terminals and credentials. Handle it the way you would any other tool with that reach. The first decision is how your policy gets onto those machines. Enterprise seats come with server-managed settings and analytics. A Console organization issues each person a key in a dedicated Claude Code workspace; you can set per-user monthly spend limits, and the key dies when the person leaves. If you run it through a cloud route behind your gateway, you push managed settings files with device management.

  • Block reads of secrets: .env files, key stores and cloud credential folders.
  • Keep a short allow-list of shell commands Claude Code can run without asking. Everything else prompts the developer.
  • Allow only MCP servers registered through the MCP gateway.
  • Block direct pushes to protected branches. Changes Claude writes go through pull requests and review like anyone else’s.
  • On networks without internet egress, point Claude Code at Bedrock through a PrivateLink endpoint, or at your gateway, and turn off nonessential traffic such as telemetry.

10. Put spend limits in three places

One runaway loop shouldn’t be able to produce a bill that surprises anyone. Set limits per app at the gateway, per workspace or account at the provider, and per month in finance. Anthropic checks workspace and organization limits on every request, so the organization limit still applies to a workspace that has none of its own.

Your model choice will save you more than your route choice. Send routine work to Claude Sonnet or Haiku and save Opus for the genuinely hard cases. Use prompt caching for long context you send again and again. And budget for agents carefully, because one task can turn into dozens of calls.

11. Mapping the controls to your frameworks

Anthropic holds ISO 27001:2022, ISO/IEC 42001:2023 and SOC 2 Type I and Type II, with reports available through its Trust Portal. That covers the provider’s side. Your side is the controls in this article, and they line up well with the frameworks regulated firms already report against. Use the mapping below as a starting point for a conversation with your auditor. It is not a certification claim.

  • Identity and agent identities: SOC 2 CC6.1 to CC6.3; ISO 27001 Annex A 5.15 to 5.18.
  • Choosing the route by data processor: SOC 2 CC9.2; ISO 27001 5.19 to 5.23 (suppliers and cloud services).
  • Redaction, classification and egress control: SOC 2 CC6.6 and CC6.7; ISO 27001 8.11, 8.12 and 8.20 to 8.22.
  • Logging and trace IDs: SOC 2 CC7.2; ISO 27001 8.15 and 8.16; EU AI Act Article 26(6) for deployers of high-risk systems.
  • Human approval: EU AI Act Article 26(2) on human oversight for high-risk systems; NIST AI RMF MAP 3.5 and GOVERN 3.2.
  • AI inventory, owners and risk assessment: SOC 2 CC3.2; ISO 27001 5.9; the ISO/IEC 42001 AI risk and impact assessment. Our AI governance operating model shows where each of these sits.

12. A realistic path to production

Build the foundation once, prove it on one use case, then bring in agents and more teams. With one cloud and one data class, plan on roughly three months to the first production use case. Each extra route or regulator adds time, and it is better to say so on day one.

  • Weeks 1 to 3, blueprint: map the use case and its data, decide the route and processor, build the threat model, and get the target design signed.
  • Weeks 4 to 8, foundation: AI gateway live, workload identity, workspaces and spend limits in place, logs flowing to the SIEM. Gate: security sign-off.
  • Weeks 9 to 14, first use case: evaluation suite, approval gates, red-team test, and a pilot with real users. Gate: the checklist below.
  • Month 4 onward, scale: MCP gateway and agents, Claude Code rollout, more data classes, and a stop drill every quarter.

Before anything goes live

Don’t put a Claude workload into production until the owner of each item below has signed it off.

  • The data processor for the route matches the DPA and the records of processing.
  • Every data class is mapped to a route, a region and a retention mode.
  • There are no static API keys in code, CI or config, and every agent has its own identity and tool list.
  • Provider endpoints can only be reached from the gateway network.
  • The gateway refuses to send traffic when authentication, classification, quotas or logging fail.
  • Every remote MCP server uses OAuth 2.1, with tokens bound to that server.
  • Writes and irreversible actions need a named approver, and someone has actually tested the kill switch.
  • The evaluation suite meets its pass marks, and red-team findings are fixed or formally accepted.
  • Logs with trace IDs reach the SIEM, and spend limits are set at the gateway, the workspace and the cloud account.

If you’d like this worked through for your own estate, we usually start with a Solution Blueprint: the route decision, the threat model and a target design your security team can sign. The rest of what we do is on our Claude implementation services page.

Common questions

Do we need an AI gateway if we only use one Claude route?

Yes. Even with one route, the gateway is the only place you can check who is calling, classify and redact data, enforce quotas and log every call. It also means you can add or switch routes later without touching application code.

How do we avoid long-lived Claude API keys?

On the Claude API, use workload identity federation: your identity provider issues a short-lived token, and Anthropic exchanges it for a short-lived token of its own. On Bedrock, Google Cloud, Foundry and Claude Platform on AWS, use cloud IAM roles. Either way, unset any old ANTHROPIC_API_KEY variable, because it overrides federation.

Can Claude traffic stay off the public internet?

On the cloud routes, yes. Use AWS PrivateLink for Bedrock and Claude Platform on AWS, Private Service Connect on Google Cloud, and a Virtual Network for the Foundry resource. The direct Claude API is reached over public TLS, so send it through an egress proxy that allows only the provider API.

Does zero data retention cover every Claude feature?

No. Batch processing, the Files API, the code execution tool, programmatic tool calling, Anthropic-hosted Managed Agents, the MCP connector, Agent Skills and the Covered Models (Claude Fable and Mythos) all keep data for a period. Put them in a separate workspace with its own approval, so a ZDR workspace can never call them by accident.

How should AI agents be controlled?

Give each agent its own identity and a short list of tools. Send every tool call through an MCP gateway with OAuth 2.1 tokens bound to one server, make writes and irreversible actions wait for a named approver, and keep a kill switch that revokes the agent and blocks its tools. Test that switch regularly.

Sources

  1. API and data retention · Claude Platform Docs
  2. Workload Identity Federation · Claude Platform Docs
  3. Workspaces · Claude Platform Docs
  4. Workspace limits (Workspaces) · Claude Platform Docs
  5. Claude in Amazon Bedrock · Claude Platform Docs
  6. Require a guardrail for model inference (IAM condition key) · Amazon Web Services
  7. What certifications has Anthropic obtained? · Claude Help Center
  8. Claude Platform on AWS · Claude Platform Docs
  9. Claude in Microsoft Foundry · Claude Platform Docs
  10. Use interface VPC endpoints (AWS PrivateLink) with Amazon Bedrock · AWS documentation
  11. Anthropic Claude on Vertex AI via a Private Service Connect endpoint · Google Codelabs
  12. VPC Service Controls with Gemini Enterprise Agent Platform · Google Cloud documentation
  13. MCP authorization (revision 2026-07-28) · Model Context Protocol
  14. Claude Code admin setup · Claude Code Docs

This article is general information, not legal or regulatory advice. Rules and products change; check the sources for the current position.

Facing this in your own estate?

Tell us what you are working on.