Evidence before adjectives
Every Critical-to-Low finding carries the observation, evidence, business risk, repair recommendation, control mapping, owner and validation method.
01 · Security exposure map
Security gap assessment, ranked by what an attacker would use first.
A read-only interrogation of configuration and evidence across 38 modules. Each finding arrives with proof, risk, ownership, a repair move and a validation test, sequenced by the route an attacker could exploit, not by alert volume.
The output
Every Critical-to-Low finding carries the observation, evidence, business risk, repair recommendation, control mapping, owner and validation method.
Priorities follow exposure, dependency and attack path, not the loudness of the alert. Identity commonly leads because every other control borrows its trust.
Your operators receive a working findings tracker; management receives a one-page view of business risk, concentration and next moves.
Assessment architecture
You select the lenses in scope. The fidelity of the map depends on the systems connected, the access granted and the evidence that exists.
Administration tiers, privileged groups, delegation, Kerberos and NTLM exposure, legacy protocols and the paths that connect them to high-value systems.
Conditional Access, identity protection, application consent, guest access and federation examined as one cloud trust fabric.
Standing administrator rights, PAM and PIM coverage, just-in-time elevation, break-glass accounts, service identities and hardened administration workstations.
MFA reach, phishing-resistant methods, legacy authentication and the policy gaps where weaker factors can still pass.
Joiner, mover and leaver breaks, orphaned identities, excessive access and segregation-of-duties collisions.
Identity lifecycle, role architecture, service identities and access certification measured for control depth and repeatability.
Anti-phishing and anti-malware controls, safe links and attachments, SPF, DKIM, DMARC and the bypass rules that can quietly nullify them.
Internet-facing services, NAT, rule hygiene, exposed administration interfaces and the change controls governing the edge.
VPN configuration, MFA, device-posture checks, split tunnelling and the practical runway toward zero-trust access.
WAF coverage, OWASP-aligned rule sets, tuning quality and the routes traffic can take around inspection.
The full ransomware path tested on paper: initial access, privilege, lateral movement, backup reach, detection and recovery time.
Coverage, immutability, separation from production credentials, restore evidence and alignment with stated RTO and RPO targets.
Plans, decision rights, investigative logging, evidence handling and readiness to rehearse the response at tabletop.
Sensor coverage, policy fractures, tamper protection, attack-surface-reduction rules and readiness to contain from the console.
Baseline alignment with applicable CIS Benchmarks, patch posture and the reach of local administrator access.
Risk-led ordering through known-exploited catalogues, EPSS, internet reachability, asset criticality and attack-path context.
Secure Score, Defender for Office 365, Purview, external sharing and audit configuration read beyond the headline score.
Subscriptions, RBAC, Azure Policy, network exposure, Key Vault, Defender for Cloud and logging against the Microsoft Cloud Security Benchmark.
IAM, organisations and SCPs, VPC exposure, S3, KMS, CloudTrail and GuardDuty against the AWS Well-Architected Security Pillar.
Resource hierarchy, IAM, VPC, storage, KMS and logging against Google Cloud Enterprise Foundations.
One view of identity, network, data and logging gaps across every cloud you run, with CSPM findings ranked by exposure.
Database access, encryption, auditability, network exposure and platform hardening examined as a single data trust boundary.
vCenter and ESXi hardening, lockdown mode, admin access, management network isolation and patch level.
Host hardening, administrative separation, shielded VMs, replication and management access.
Prism access, cluster hardening, STIG alignment, encryption and management-plane isolation.
NetScaler exposure and patch level, delivery controller access, session policies and the gateway as an internet-facing door.
East-west policy, inter-zone rules and least privilege between segments.
Zones, micro-segmentation, VLAN and ACL design, and whether trust boundaries exist in configuration as clearly as on diagrams.
802.1X, guest and BYOD isolation, admission policy and wireless security at the point a device joins.
Classification, DLP policy, encryption and key management traced from declared sensitivity to enforced behaviour.
Log-source coverage, use-case quality, alert fidelity, SOAR, retention and ingestion economics assessed as one detection system.
Whether the systems that matter are actually generating the audit events an investigation would need.
Every expected source confirmed as arriving in the SIEM, on time, from every in-scope host.
Fields parsed, normalised and usable by detections, not stored as unreadable text.
MITRE ATT&CK coverage, rule logic, tuning discipline and test evidence, because a detection that never fires is only documentation.
Agreed, harmless test events run in a controlled window to confirm that priority detections fire and reach an analyst.
Hypothesis-led, ATT&CK-mapped hunts across existing telemetry, with what is learned engineered back into durable detections.
Retention against investigation and regulatory needs, and ingestion spend against detection value.
Access geometry
| Permission band | Purpose | Grant pattern |
|---|---|---|
| ROObservation | Every assessment lens; configuration remains untouched. | Your administrator grants reader or security-reader roles, or supplies configuration exports, for the defined assessment window. |
| EDDocument edit | Named policy, standard and audit-document revisions. | Edit rights apply only to identified documents, with tracked changes and your approval retained in the record. |
| RWIntervention | Remediation only, after you approve both the finding and the change plan. | A separate, time-boxed administrator grant, such as PIM activation, for the named systems and approved change window. |
No agent is installed on your systems: FlowCraft AI runs on our side and works through the access you grant. Your data stays on your systems. Timing follows the selected scope and is fixed in the written proposal; the clock begins only after access is confirmed. How remote delivery works.
Framework translation
Identity, email and recovery usually reveal the first decisive moves. Tell us what you operate and where confidence thins; we will propose the right assessment boundary.