Complete your evidence coverage

Sign in, connect AWS and/or Microsoft, then generate the posture report. Aevyra uses the same finding rules and interactive report for live and demo evidence. Live results depend on what the connected services and your applications actually expose.

Enable the additional Microsoft checks

1. Open your connections and select Enable Microsoft security checks (or Connect Microsoft).
2. Have an authorized Microsoft administrator approve the delegated read-only permissions in Microsoft's sign-in window. A working policy connection remains active until the upgrade succeeds.
3. Generate a fresh report and review Microsoft source coverage. Each source reports missing consent, role or licensing separately.

Extra delegated permissions: SecurityAlert.Read.All, SecurityEvents.Read.All, IdentityRiskEvent.Read.All, AuditLog.Read.All, DeviceManagementManagedDevices.Read.All. A signed-in user also needs the source's supported reader role and the relevant service license. MFA registration reporting supports Reports Reader or Security Reader; security alerts and identity risk reporting require supported security reader roles. Intune requires an active Intune license. Existing Purview scopes remain in place. This update does not purchase licenses, enable paid services or alter customer policies.

Microsoft checks cover Defender alerts, identity risk detections, MFA registration and recent device compliance/encryption where available. Registration does not prove MFA enforcement. Secure Score is contextual: recommendations without provider severity remain unscored. Purview applicable policies establish scope, not a claim that policies prevented every leak. Security events and identity risk detections use a 30-day window.

Review AWS read access

On an existing AWS connection, expand Review AWS read-only permissions. Your AWS administrator can copy the updated policy onto the existing evidence role. It adds reads for key lifecycle, tags, Lambda, Inspector, Config, service TLS and existing Bedrock log events. Collection remains limited to the connected account and supported regions. The update does not enable Inspector, Security Hub, Config or logging, invoke AI models, rotate keys or change resources.

Existing cloud findings, certificates, encryption settings and TLS bindings flow into the report. Optional sources have bounded reads; access denied, unavailable services and limits appear as coverage gaps. AWS Lambda.2 findings provide a supported runtime signal; Aevyra does not guess every runtime's support status from its version string.

Application and governance facts

Cloud inventory cannot establish your approved AI tools, data retention requirements, HSM target compatibility, exceptions, ownership or approved baselines on its own. To assess those findings, export facts from an authorized application, policy register or local Collector.

1. Download the JSON template and the JSON schema.
2. Replace placeholder identities and references with your own stable records. Set only facts the source establishes. Leave unknown values as null for posture facts or "unknown" for AI facts. Omit unused channels and subjects.
3. In your connections, expand Include application, AI activity or governance evidence (optional) and select the JSON file. Generate the report to include it in the same sealed assessment as AWS and Microsoft. A correction requires a fresh report once the run is sealed.

The Python Collector source also accepts --facts-file /path/to/facts.json on its existing connect command. Browser upload is available immediately after this update; previously downloaded native Collector binaries need a normal rebuilt release to include the new option. A customer export remains labelled customer-confirmed. System allowances still apply to distinct subjects.

Each subject has system_id, optional system and business_unit, and a nonempty facts array. Every fact has fact, value, provenance and a bounded source_id; observed_at is optional. Provenance values: collector_observed, connector_observed, customer_confirmed or imported. Browser exports are relabelled customer-confirmed. Repeated or conflicting observations cannot silently become a pass.

Use references and facts only: no prompts, documents, credentials, raw logs, assessment scores, severity or recommended actions. Maximum file size: 512 KB; 1,000 subjects per channel. Use the same stable system ID across inventory and facts to join the observations.

Posture facts

minimum_rsa_bits must be an integer from 1024 to 16384 representing an approved minimum. Date-only values mean midnight UTC. Numeric intervals are days or years as named. HSM compatibility needs hsm_target_profile; policy compliance needs policy_id; drift needs baseline_id; vendor dependency needs vendor. Empty values do not establish those checks.

Every cryptographic and governance demo finding

Every AI demo finding

AI facts use true, false or unknown. Approvals must come from your policy owner; provider configuration, model presence and pattern matches cannot establish approved use or confirmed exfiltration. The engine derives severity from the supplied combination of facts.

Existing authorized Bedrock samples can produce a Medium review finding for potential secret markers. A candidate match needs validation; it is not a confirmed data leak. An empty sample does not establish safe use. Confirmed behaviors need explicit supported evidence in the factual feed.

Read the result

Finding detection coverage lists all 28 demo finding types and the evidence still needed. It is separate from resource/check coverage. A source completing with no rows does not prove that all controls passed. Missing grants, stale devices, unsupported services and conflicting facts remain unknown. Every recorded finding participates in the shared explorer, filters and supported scores; informational recommendations remain unscored where severity is unknown.

Source requirements

• Microsoft security alerts
• Microsoft identity risk detections
• Microsoft MFA registration report
• Microsoft Intune managed devices
• AWS Lambda security controls