MCP covers every Growth and Scale feature

MCP covers every Growth and Scale feature

Plans: every plan — the tools you see follow the plan you are on (backend 2.5.13). Where: the /mcp endpoint; user menu → API Tokens; the MCP Tool Catalog

Until now an AI client connected over MCP could read access intelligence, ISO 27001 basics, risks, suppliers and the audit trail — 30 tools — regardless of which plan your organisation was on. This release does two things at once: it adds MCP tools for every Growth and Scale feature the console already has, and it makes the tool list honour your plan, exactly as the console does.

What’s new

34 new tools — 64 in total, 59 read-only

Every tool is organisation-scoped, PII-masked and returns a projected result — no free text over 500 characters, no repository coordinates, no contacts, no other organisation’s review comments. The new families, with the lowest plan that includes them (Growth means The Growth, Scale means The Scale):

• Access (Growth) — access_zombie_accounts, access_toxic_combinations.
• ISO 27001 (Growth) — iso_roadmap_list, iso_policies_list, iso_readiness_score, iso_evidence_list; certification bodies (Growth) — iso_cb_directory_list, iso_cb_profile_get; certificate status (Scale) — iso_certificate_status.
• Compliance (Growth) — compliance_posture, compliance_coverage_matrix, compliance_objectives_list, compliance_crosswalk_lookup, compliance_shared_evidence.
• Continuous monitoring (Scale) — monitoring_status, compliance_control_reliability, compliance_exceptions_list; HIPAA (Scale) — compliance_hipaa_readiness; board reporting (Scale) — board_summary.
• Automated evidence (Growth) — evidence_jobs_list, evidence_job_get.
• Supply chain security (Growth) — sscs_sbom_list, sscs_sbom_components, sscs_findings_list, sscs_findings_summary, sscs_vendor_list, sscs_vendor_risk_summary.
• Privacy operations (Scale, metadata only) — privacy_cockpit, privacy_activities_list, privacy_activity_get, privacy_dpia_list, privacy_dpia_get, privacy_data_dictionary_list.
• Audit rehearsal (Scale, mcp:write) — iso_audit_simulate puts one question to a simulated auditor and returns model-generated rehearsal text. It is a write tool, rate-limited per token, and — like every write tool — stays out of tools/list until MCP writes are enabled for your deployment.

See MCP Capabilities for the map and the Tool Catalog for every argument and result shape.

Tools follow your plan

Each tool is tied to a feature, and features belong to plans. A token from an organisation on a lower plan does not see the tool in tools/list and cannot call it; a call by name returns the SDK’s unknown-tool error. Upgrade and the same token sees the new tools on its next tools/list — nothing to re-issue. Scope is not entitlement: mcp:write on a token from an organisation on The Seed plan still unlocks no Growth tool. See Scope is not entitlement.

Two tools stay on every plan: maturity_scores and posture_summary (and employee_resolve for a pii:read token). Seven of the existing access reports need Seed; audit_trail_search, uar_status, the three advanced access reports and the ISO tools need Growth — the same plans that show those pages in the console.

In user menu → API Tokens → Create Token, ticking MCP Read now shows a one-line note naming your plan, with a See what each plan includes link to Billing & Subscription.

A “Requires plan” column in the Tool Catalog

The generated tool table gains a Requires plan column after Scope — All plans, Seed and above, Growth and above or Scale — derived at build time from the same pricing configuration the console uses. The published MCP manifest carries the underlying feature ID as requiresFeature on every tool, and both SDKs expose it.

Changes to existing tools

• audit_trail_search returns projected entries — { seqNum, createdAt, actorId, action { code, group }, subjectType, subjectId } — inside { items, total, truncated }. Names, change payloads and chain fields are no longer included, and it now needs the Audit Trail feature (Growth and above). New filters: subjectType and action.
• access_reviews_list returns { items, total } of campaign summaries instead of a bare array, and no longer includes reviewer names. Read .items.
• Name keys. MCP masks the generic name key on every result. Tools whose “name” is not a person’s return it as sbomName, componentName, vendorName, activityName, entryName or bodyName.

For SDK users

• List conventions. Every list tool takes limit (default 50, maximum 200) and offset, and returns total plus truncated: true when there is more.
• Two new non-retryable errors. NOT_PERMITTED can now carry reason: "FEATURE_NOT_ENTITLED" and a feature ID when a tool registration outlives a plan change — refresh tools/list. RESULT_TOO_LARGE replaces any result over 256 KiB and carries the limit in bytes — narrow the request. Both SDKs raise them as non-retryable ToolErrors; neither is retried. See MCP Errors and Retries.
• Regenerated TypeScript and Python clients cover all 64 tools.