TechOneDigital Review one MCP trust boundary

Based on the current stable MCP specification

Security & authorisation

MCP security is the design of authority, not the addition of an OAuth screen.

MCP standardises how AI applications discover and invoke capabilities. Trust, data access, business authorisation and approval remain architecture decisions enforced from the host to the system of record.

For

Security architects, IAM leads and system owners

Use when

MCP can expose sensitive data or consequential actions.

Outcome

An authority model, enforceable controls and an audit trail for every consequential action.

Technical guide by , Founder & Technology Consultant.

TechOne architecture position

The trust boundary is the exact principal, server, capability, data class and consequence involved in a request. Preserve it through the host, gateway, MCP server and system of record.

Illustrative architecture scenario, not a customer case.

See these boundaries in one CRM change.

Follow a proposed opportunity update through identity, object permissions, exact-action approval, command execution and outcome reconciliation. The example connects gateway, security and production decisions.

Follow the CRM walkthrough

Three different decisions

Authentication, authorisation and approval are not synonyms.

An identity can be valid, permitted to use a capability and still require a human decision for one exact change. Treating those checks as one boolean creates invisible authority.

Security decision boundaries

Authentication

Authentication establishes who or what is calling.

Protocol basisFor protected HTTP deployments, the current MCP specification defines an OAuth-based model in which the MCP server acts as a resource server. Access tokens must be audience-bound to that MCP server, carried in the Authorization header and validated on every protected request.TechOne recommendationRepresent the human user, workload or delegated agent explicitly. Derive principal and tenant from verified identity claims, not from model-supplied tool arguments. Prefer short-lived credentials and managed workload identities over shared secrets.

Authorization

Authorisation establishes what that principal may do.

Protocol basisThe available tools and resources may vary according to the authorisation supplied on each request. MCP supports least-privilege scope challenges and step-up authorisation. An MCP server must not accept a token intended for another resource or pass its inbound token through to a downstream API.TechOne recommendationEnforce both capability-level and business-object authorisation. A tools/call permission is not evidence that the caller may update every invoice, case, patient record or production environment reachable through that tool.

Approval

Approval establishes whether this action is intended now.

Protocol basisThe tools specification recommends a human in the loop with the ability to deny invocations, clear visibility of tool inputs and confirmation for sensitive operations. The protocol does not prescribe one universal approval user experience.TechOne recommendationRequire approval according to consequence, not according to whether a tool is labelled read or write. Show the target, material parameters, data leaving the trust boundary and expected side effect. Bind an approval to the exact request so that later parameter changes require a new decision.

Threat → control → evidence

Security claims should be inspectable.

The table separates protocol requirements and documented protocol threats from TechOne architecture recommendations.

RiskControlRelease evidenceOwnerBasis
Prompt injection through resources, tool results or tool metadataTreat retrieved content, descriptions and results as untrusted input. Minimise the tools available in the same context, separate sensitive reads from consequential writes, validate outputs and require explicit approval for material actions.The MCP specification treats tool annotations as untrusted unless they come from a trusted server. Official OpenAI and Microsoft guidance identifies prompt injection through connected content, tool descriptions and tool outputs as a production risk.AI application owner with security engineeringProtocol guidance; TechOne recommendation
Tool poisoning or a post-approval schema changeRecord a known-good tool definition, publisher, endpoint and version. Diff names, descriptions, input schemas, output schemas and annotations before release; require re-approval for material changes.Microsoft advises treating tool schema changes as dependency changes and pinning known-good schemas. OpenAI notes that an MCP server may change tool behaviour unexpectedly.MCP platform ownerOfficial platform guidance; TechOne recommendation
Name collision or tool shadowing across serversUse an immutable internal server identifier and a namespaced capability identifier in policy, telemetry and approval records. Never authorise solely by the display name supplied to the model.The MCP specification scopes tool-name uniqueness to one server, warns aggregators about collisions and states that serverInfo.name is not guaranteed to be unique.Host and gateway engineeringProtocol constraint; TechOne recommendation
Confused deputy and excess server privilegeSeparate the server’s execution identity from caller authorisation. Check the caller on every operation and issue a separate, correctly scoped downstream credential when the server calls another protected API.The MCP authorisation security specification describes the confused-deputy problem and requires per-client consent in affected proxy flows. It explicitly prohibits token passthrough.Identity architect and service ownerProtocol requirement and guidance; TechOne recommendation
Token theft, replay or cross-resource token useValidate issuer, audience, expiry and scopes; use PKCE for authorisation-code flows; protect tokens in storage and logs; rotate refresh tokens for public clients and prefer short-lived access tokens.These controls are normative requirements or recommendations in the current MCP authorisation and security specifications.Identity platform ownerProtocol requirements and guidance
SSRF during OAuth discovery, CIMD retrieval or URL handlingRequire HTTPS in production, validate each redirect, block private, loopback and link-local destinations where they are not explicitly required, apply egress policy and avoid home-grown IP parsing.The official MCP security guidance documents SSRF paths through resource metadata, authorisation-server metadata and Client ID Metadata Documents, including redirects and DNS rebinding.Network and identity securityProtocol guidance; TechOne recommendation
DNS rebinding against a local HTTP MCP serverValidate Origin on Streamable HTTP, bind local servers to loopback and require authentication. Prefer stdio or access-controlled IPC where a local HTTP listener is unnecessary.The Streamable HTTP specification requires Origin validation and recommends loopback binding and authentication for local deployments.MCP server ownerProtocol requirement and guidance; TechOne recommendation
State-handle hijackingUse opaque, high-entropy, expiring handles and bind each handle server-side to the verified principal and tenant. Possession of a handle must not count as authentication.The current MCP specification has no protocol sessions and documents explicit application-state handles. Official guidance advises checking the caller’s authorisation on every use and recommends principal binding and non-deterministic handles.MCP server ownerProtocol guidance
Local stdio process compromiseAllow only reviewed launch commands, pass a minimal environment, sandbox filesystem and network access, and run with the least operating-system privilege required.Official MCP security guidance treats local servers as executable code with the privileges of the client and recommends explicit consent, sandboxing and restricted resources.Endpoint security and developer platformProtocol guidance; TechOne recommendation
Private data served from a shared cacheUse cacheScope private whenever a result depends on identity, scopes or tenant, and partition private caches by a stable authorisation context. Continue to enforce access control on the underlying primitive.The MCP caching specification prohibits sharing private results across authorisation contexts and states that cacheScope is not an access-control mechanism.Client, gateway and cache ownerProtocol requirement; TechOne recommendation
Header and body disagreement at an intermediaryReject requests when the required MCP-Protocol-Version and Mcp-Method headers, any applicable Mcp-Name header, or recognised Mcp-Param values are missing, malformed or do not match the JSON-RPC body.The current Streamable HTTP specification mandates header/body validation and defines HeaderMismatch error -32020.Gateway and MCP server ownerProtocol requirement
Approval replay or parameter substitution after consentBind approval to the verified principal, server-qualified capability, material parameters, target, environment, schema version and expiry. Canonicalise that request, record its digest and reject any replay or invocation whose bound values differ; material changes require a new approval.The MCP tools specification recommends clear visibility of tool inputs and human confirmation for sensitive operations. Exact request binding, expiry and replay prevention are TechOne controls that make the approval inspectable and enforceable.Host, approval service and policy ownerProtocol guidance; TechOne recommendation

Control point, not a safety proof

A gateway is a control point, not a complete safety system.

The current Streamable HTTP transport makes MCP easier to govern with ordinary HTTP infrastructure. That is valuable, but the gateway sees only part of the decision.

Useful central controls

  • Validate caller identity, token audience and transport requirements.
  • Allow or deny endpoints, protocol versions, methods and named capabilities.
  • Apply rate, concurrency, network, regional and cost policies.
  • Route on Mcp-Method and Mcp-Name without parsing the JSON body, while requiring the server to validate the mirrored values.
  • Attach correlation identifiers and produce consistent access and policy logs.
  • Protect downstream services with timeouts, quotas and circuit-breaking policy.

Remaining trust questions

  • Prove that retrieved content contains no prompt injection.
  • Decide whether a user genuinely intended a specific business consequence.
  • Replace object-level authorisation inside the system of record.
  • Make an unsafe or misleading tool description trustworthy.
  • Provide business idempotency for a payment, order or deployment.
  • Turn a self-reported registry entry into verified software provenance.

Illustrative evidence record

Record the decision chain, not merely the HTTP request.

MCP reserves W3C trace context keys for distributed tracing. A defensible audit record must add the identity, policy, approval and business context needed to reconstruct why an action was allowed.

trace_id and request_id
Correlate host, gateway, MCP server and downstream spans.
Protocol-supported and TechOne recommendation
principal, workload and tenant
Identify the verified human and non-human actors without trusting tool arguments.
TechOne recommendation
server ID, endpoint and deployed version
Distinguish the actual implementation from its display name.
TechOne recommendation
tool name and schema fingerprint
Show which reviewed capability contract was invoked.
TechOne recommendation
scopes and policy decision
Explain the authorisation basis and the rule that allowed or denied the call.
TechOne recommendation
approval decision and bound request digest
Show exactly what a person approved and prevent approval reuse after parameter changes.
TechOne recommendation
target class and redacted argument digest
Support investigation without copying secrets or regulated payloads into general logs.
TechOne recommendation
outcome, error class and latency
Connect security evidence with operational reliability.
TechOne recommendation

TechOne security position

Design for refusal, revocation and investigation.

  1. 01

    Authorise a capability, then authorise the object.

    Permission to call update_invoice is not permission to update every invoice reachable by the server.

    TechOne recommendation
  2. 02

    Trust is contextual, not inherited.

    A trusted publisher does not make all content returned through that server trustworthy, and a trusted read tool can still expose sensitive data.

    TechOne recommendation
  3. 03

    Use separate credentials at separate resource boundaries.

    Audience-bound MCP access and downstream API access are distinct security decisions.

    Protocol fact
  4. 04

    Make consequential authority scarce.

    Expose the smallest set of write operations, parameters, records and environments required for the workflow.

    TechOne recommendation
  5. 05

    Treat metadata as code-adjacent supply chain.

    Names, descriptions, schemas and annotations influence model behaviour and require review and change control.

    TechOne recommendation
  6. 06

    Fail closed at ambiguous boundaries.

    Unknown issuer, audience, endpoint, schema version, tenant or approval state should stop the action rather than trigger a permissive fallback.

    TechOne recommendation

For the agent behind the MCP client, see AI agent identity, guardrails and a kill switch.

Commercial route

Review one MCP trust boundary before it reaches production.

We trace one proposed or existing capability from host and identity through any gateway and MCP server to the system of record. The output is a prioritised set of authorisation, approval, data-handling and evidence controls for that boundary.

Review one MCP trust boundaryExplore the complete MCP architecture

Primary sources

Protocol claims are grounded in official sources.

The source list distinguishes core protocol requirements, optional extensions and external implementation guidance. The surrounding enterprise control model is a TechOne architecture recommendation.

  1. MCP specificationNormative protocol baseline
  2. ArchitectureHost, client and server responsibilities
  3. MCP specification changelogTrace context, lifecycle and protocol changes
  4. Streamable HTTPTransport, Origin and header validation
  5. AuthorisationOAuth discovery, scopes and token use
  6. Authorisation security considerationsAudience binding, token passthrough, SSRF and confused deputy
  7. Security best practicesImplementation threats and mitigations
  8. ToolsTool control, schemas, collisions and safety
  9. CachingCache scope and authorisation boundaries
  10. Official MCP Registry trust modelNamespace ownership, publisher metadata and security-scanning boundary
  11. MCP servers: risks and safetyPrompt injection, approvals and third-party servers
  12. Secure Azure MCP Server deploymentTool poisoning, gateway and operational controls

Direct answers

Questions that expose the hidden architecture decision.

Does MCP require OAuth for every server?

No. Authorisation is optional in the core protocol. When a protected server uses an HTTP transport, the current MCP specification defines an OAuth-based authorisation model. Local stdio deployments use a different process and credential boundary.

Is OAuth enough to secure an MCP deployment?

No. OAuth establishes and scopes access to the MCP resource. The server must still enforce business-object permissions, validate inputs, constrain downstream authority, protect outputs and apply approval policy to consequential operations.

Can an MCP gateway stop prompt injection?

Not by itself. A gateway can constrain traffic, identity, methods and destinations. Prompt injection can arrive inside content already permitted by those controls, so the host and application need context isolation, tool filtering, validation and consequence-aware approval.

Is a read-only MCP server safe?

Read-only reduces direct mutation risk, but it can still expose sensitive data, carry malicious instructions or provide information that another tool later exfiltrates. Read authority must still be scoped and audited.

May an MCP server pass the caller’s token to its backend API?

No. MCP explicitly forbids token passthrough. The inbound token must be intended for the MCP server; access to an upstream API requires a separate token issued for that upstream resource.

Does the official MCP Registry prove that a server is safe?

No. The official registry verifies that a publisher controls the GitHub or domain namespace used for publication and hosts publisher-maintained metadata. It does not audit the server code, package, deployment or runtime behaviour. An enterprise still needs allowlisting, review, version control and runtime policy.

Does a confirmation dialog count as business approval?

Not by itself. A confirmation prompt can support user intent, but an enforceable approval must identify the approver and bind the decision to the exact server-qualified capability, material parameters, target, environment and validity window. A changed or replayed request requires a new decision.

What matters in the current MCP security model?

Requests are self-contained, there are no protocol sessions, method and name headers can support routing, issuer and audience handling are stricter, and Client ID Metadata Documents replace Dynamic Client Registration in the current model. Application state handles also have explicit security guidance. Existing deployments still need compatibility testing with older clients and servers.